--- schema_version: 1 rubric_version: "1.1" kind: pair canonical_url: "https://onaihq.com/compare/agents/meingpt-vs-t3-chat/" data_url: "https://onaihq.com/compare/agents/meingpt-vs-t3-chat/data.json" generated_at: 2026-10-27 last_checked_at: 2026-10-26 last_cross_checked_at: 2026-10-26 next_check_at: 2026-12-20 publisher: "onAI" --- # meinGPT vs T3 Chat > Published by onAI. onAI also sells one of the products compared here. Findings about other products link to public sources so you can check them; statements about onAI are marked as stated by onAI. This comparison has 26 criteria with sourced findings for both products, including 26 that a second agent cross-checked for both. 49 criteria have a sourced finding for at least one product. "Sourced" findings passed publication validation: they are scoped, fresh and cite public resolved sources. "Agent cross-checked" findings also received a separate agent source pass, not necessarily a hands-on test. "Stated by onAI" marks the publisher's own statements about its product (evidence type vendor-statement), which cite no third-party source. Humans approve the rubric, templates and generator; they do not review every finding. It does not declare a winner. A missing sourced finding is not evidence that a capability is absent. - meinGPT source checks: 2026-10-21 to 2026-10-26. - T3 Chat source checks: 2026-10-21. - Canonical HTML: https://onaihq.com/compare/agents/meingpt-vs-t3-chat/ - Machine-readable JSON: https://onaihq.com/compare/agents/meingpt-vs-t3-chat/data.json - Methodology: https://onaihq.com/compare/agents/methodology/ ## Models ### MOD-01: Model catalogue and providers Judgment rule: List each available model by provider and plan, including reasoning, fast and multimodal variants. Record whether the user picks the model, the product routes automatically, or both. A provider counts only if at least one of its current models is selectable on the scoped plan. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial Curated multi-provider models are selected per chat or by context-length routing; the signed DPA and admin approvals affect availability. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Business and individually contracted Enterprise web workspace; permitted models depend on workspace approval and DPA; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - auto_routing: Bestes Modell starts with GPT-5 and changes to Gemini 3.5 Flash-Lite only when context is too long. This is length routing, not task routing; visibility of the final routed identity is not established. - providers — Which model providers are available? [supported]: Model authors include OpenAI, Anthropic, Google, Mistral, DeepSeek, Moonshot, Qwen, GLM and MiniMax. Hosts are separate from authors. Sources: meingpt:meingpt:models-rev2, meingpt:meingpt:lifecycle-rev2. - current_models — Which current models are selectable on this plan? [supported]: Active table lists GPT-4.1, GPT-4o/Mini, GPT-5/mini/Nano/Thinking/5.2/5.4/5.5/5.6 Luna/Sol/Terra, GPT-6 Luna/Sol/Astra, GPT-6.1 Sol; o3; Claude Haiku 5.5, Opus 4.5/4.6/4.7/4.8/5/5.5, Sonnet 4.6/5/5.5; Gemini 3.5 Flash/Flash-Lite, 3.6/3.7/3.8 Flash and 4 Argon; MeinGPT Web; Mistral Medium 3.5/CodeStral; GPT-OSS 120B; Kimi K2.6/K3; DeepSeek V3.2/V4 Flash/V4 Pro/V4.1 Flash; GLM-5.3/Flash, MiniMax-M3 and Qwen3.8 Max. Duplicate EU/provider routes exist. GPT-6 Astra requires accepted AVV 2.0. Model labels do not guarantee one-to-one public provider variant names. Sources: meingpt:meingpt:lifecycle-rev2, meingpt:meingpt:models-rev2, meingpt:meingpt:rates. - user_choice — Can users pick the model per chat? [supported]: Users select and switch a permitted model per chat. Sources: meingpt:meingpt:models-rev2. - auto_routing — Does the product route between models automatically, and is the chosen model shown? [partial]: Bestes Modell starts with GPT-5 and changes to Gemini 3.5 Flash-Lite only when context is too long. This is length routing, not task routing; visibility of the final routed identity is not established. Sources: meingpt:meingpt:models-rev2. - multimodal_input — Which models accept images, audio or files? [supported]: GPT-4o/5, Claude, Gemini and Mistral Medium/Magistral accept JPEG/PNG/GIF/WEBP; DeepSeek, GPT-OSS, CodeStral and older Kimi K2 do not. Audio uses separate transcription. Sources: meingpt:meingpt:models-rev2. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:models-rev2, meingpt:meingpt:lifecycle-rev2, meingpt:meingpt:rates #### T3 Chat Verification status: Agent cross-checked (agent-cross-checked) Support: partial Paid-account vendor replies establish manual model switching and provider routes, while the current plan-specific catalogue is unknown. Availability: generally-available Delivery: native Scope: surfaces=T3 Chat browser application; version=Public FAQ, legal pages and vendor replies retrieved 2026-10-21; no authenticated deployment observed; plan=Individual Pro USD8 vendor-support context; no Free or Premier catalogue parity asserted; add_ons=Optional customer API keys; plan-specific quotas and checkout details not fully public; deployment_mode=not specified; region=US-hosted service with provider processing in US or other countries; no EU/CH-only option established; os=not specified. Limitations: - Documented findings only; unestablished atomic questions remain explicit gaps. - No complete Free/Pro/Premier entitlement matrix or authenticated model/feature menu was retrieved. - providers — Which model providers are available? [partial]: Pro-context vendor replies discuss Gemini, Claude and GPT routes; the FAQ separately describes OpenAI/Anthropic/Google/OpenRouter data handling. Those policy/provider names are not a complete tier-entitlement or underlying-lab catalogue. Sources: t3-chat:t3-chat:billing-change, t3-chat:t3-chat:faq-training, t3-chat:t3-chat:r2-paid-tier-controls. - current_models — Which current models are selectable on this plan? [unknown]: A complete current selectable-model list for Pro, Free or Premier is unverified. The FAQ claude-4.6-sonnet URL is an example, and undated support model names are not a current plan catalogue. - user_choice — Can users pick the model per chat? [partial]: In the USD8 context discussion the vendor recommends manually switching to Gemini Flash or summarizing into a new chat with another chosen model. This supports Pro model selection as described there, but not a current all-model, effort or Free/Premier entitlement matrix. Sources: t3-chat:t3-chat:r2-pro-context-byok. - auto_routing — Does the product route between models automatically, and is the chosen model shown? [unknown]: Does the product route between models automatically, and is the chosen model shown? The public T3 FAQ, Terms, Privacy/Security policies, official feedback statuses and identifiable vendor replies did not establish this detail. No authenticated account, enterprise quote or bounded fixture was used. - multimodal_input — Which models accept images, audio or files? [unknown]: No current Pro model-by-model image/audio/file compatibility matrix was retrieved. The generic privacy upload categories do not establish a plan entitlement. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: t3-chat:t3-chat:billing-change, t3-chat:t3-chat:faq-training, t3-chat:t3-chat:r2-paid-tier-controls, t3-chat:t3-chat:r2-pro-context-byok ### MOD-02: Model freshness Judgment rule: Measure the days from a provider's general release to availability in the product for the last three major releases from OpenAI, Anthropic and Google (and Mistral where offered). Use dated changelogs or observation. Record deprecation notice for retired models. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial A dated changelog and explicit retirement rules exist; comparative release-lag medians were not established. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - version_pinning: An available model can be selected, but retirement automatically substitutes a successor in assistants, workflows and chats; indefinite retention of a retired model is excluded. - Unverified release_lag_days: Provider release dates for the last three major releases per provider were not reconciled against product release dates, so no lag or median is asserted. - release_lag_days — What was the lag for each of the last three major releases, per provider? [unknown]: What was the lag for each of the last three major releases, per provider? Unknown. Provider release dates for the last three major releases per provider were not reconciled against product release dates, so no lag or median is asserted. - changelog — Is there a dated public model changelog? [supported]: Public releases include v2.67.0 dated 2026-10-18. Sources: meingpt:meingpt:changes. - deprecation_notice — How much notice is given before a model is removed? [supported]: Normally four weeks, then reminders at two and one week; provider short notice can shorten this. Opus 4.5 is listed for 23 November retirement; other future dates are in the table. Sources: meingpt:meingpt:lifecycle. - version_pinning — Can an admin keep an older model version? [partial]: An available model can be selected, but retirement automatically substitutes a successor in assistants, workflows and chats; indefinite retention of a retired model is excluded. Sources: meingpt:meingpt:lifecycle. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:changes, meingpt:meingpt:lifecycle #### T3 Chat Verification status: Agent cross-checked (agent-cross-checked) Support: partial Public feature-release information is described; plan-specific pinning and version guarantees remain unknown. Availability: generally-available Delivery: native Scope: surfaces=T3 Chat browser application; version=Public FAQ, legal pages and vendor replies retrieved 2026-10-21; no authenticated deployment observed; plan=Public T3 Chat release information; no verified Free/Pro/Premier picker/pinning entitlement; add_ons=Optional customer API keys; plan-specific quotas and checkout details not fully public; deployment_mode=not specified; region=US-hosted service with provider processing in US or other countries; no EU/CH-only option established; os=not specified. Limitations: - Documented findings only; unestablished atomic questions remain explicit gaps. - No complete Free/Pro/Premier entitlement matrix or authenticated model/feature menu was retrieved. - release_lag_days — What was the lag for each of the last three major releases, per provider? [unknown]: What was the lag for each of the last three major releases, per provider? No matched dates for the last three provider releases, fixed removal notice or contractual model-version retention was retrieved. - changelog — Is there a dated public model changelog? [partial]: A vendor reply says major-feature changelog lives in Discord and a website version is planned. No accessible dated model-by-model release ledger was retrieved. Sources: t3-chat:t3-chat:changelog. - deprecation_notice — How much notice is given before a model is removed? [unknown]: How much notice is given before a model is removed? No matched dates for the last three provider releases, fixed removal notice or contractual model-version retention was retrieved. - version_pinning — Can an admin keep an older model version? [unknown]: The vendor describes pinning and a Legacy designation in the general picker discussion, without a named individual tier. Plan-specific pinning, retained model versions and admin locks remain unverified. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: t3-chat:t3-chat:changelog ### MOD-03: Bring-your-own keys and provider contracts Judgment rule: Record customer-held API keys, customer cloud accounts (Azure OpenAI, Bedrock, Vertex), and aggregator routes (OpenRouter, LiteLLM) as separate routes. Record what still works without a vendor-provided model, whether prices change with BYO, and documented restrictions. API-key support does not imply support for the customer's own Azure contract. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial Customer LLM API keys, cloud contracts and arbitrary model endpoints are explicitly excluded from the hosted catalogue. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - own_api_keys: The hosted workspace does not accept customer OpenAI, Azure OpenAI or Anthropic keys. DeepL translation BYOK is separate. - own_cloud_contract: LLM requests use meinGPT-managed provider accounts, not customer Azure/OpenAI contracts. - custom_endpoint: Admins cannot add a model or provider key to the catalogue. Custom MCP adds tools, not LLM endpoints. - byo_pricing: There is no hosted LLM BYOK licence schedule because LLM BYOK is excluded. DeepL BYOK bills translation directly through DeepL. - byo_restrictions: Only DeepL has a documented BYOK exception; it does not replace the workspace model route. - own_api_keys — Can the customer use its own provider API keys? [unsupported]: The hosted workspace does not accept customer OpenAI, Azure OpenAI or Anthropic keys. DeepL translation BYOK is separate. Sources: meingpt:meingpt:models. - own_cloud_contract — Can it use its own Azure OpenAI, Bedrock or Vertex deployment? [unsupported]: LLM requests use meinGPT-managed provider accounts, not customer Azure/OpenAI contracts. Sources: meingpt:meingpt:models. - custom_endpoint — Can an admin add any OpenAI-compatible endpoint? [unsupported]: Admins cannot add a model or provider key to the catalogue. Custom MCP adds tools, not LLM endpoints. Sources: meingpt:meingpt:models. - byo_pricing — Does BYO change the licence price? [partial]: There is no hosted LLM BYOK licence schedule because LLM BYOK is excluded. DeepL BYOK bills translation directly through DeepL. Sources: meingpt:meingpt:models, meingpt:meingpt:translation. - byo_restrictions — Which features stop working with BYO? [partial]: Only DeepL has a documented BYOK exception; it does not replace the workspace model route. Sources: meingpt:meingpt:models. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:models, meingpt:meingpt:translation #### T3 Chat Verification status: Agent cross-checked (agent-cross-checked) Support: partial Vendor replies support a Pro own-OpenRouter-key route, without establishing universal provider/model coverage. Availability: generally-available Delivery: native Scope: surfaces=T3 Chat browser application; version=Public FAQ, legal pages and vendor replies retrieved 2026-10-21; no authenticated deployment observed; plan=Individual Pro USD8 with optional customer OpenRouter key; no Free BYOK entitlement asserted; add_ons=Customer-paid OpenRouter/provider usage; model routing and verification prerequisites apply; deployment_mode=not specified; region=US-hosted service with provider processing in US or other countries; no EU/CH-only option established; os=not specified. Limitations: - Documented findings only; unestablished atomic questions remain explicit gaps. - No complete Free/Pro/Premier entitlement matrix or authenticated model/feature menu was retrieved. - own_api_keys — Can the customer use its own provider API keys? [supported]: In the paid USD8 usage discussion, staff explicitly offers a customer OpenRouter key for direct request billing; the USD8 context discussion describes OpenRouter fallback. No all-provider/all-model BYOK entitlement follows. Sources: t3-chat:t3-chat:r2-paid-tier-controls, t3-chat:t3-chat:r2-pro-context-byok. - own_cloud_contract — Can it use its own Azure OpenAI, Bedrock or Vertex deployment? [unknown]: Can it use its own Azure OpenAI, Bedrock or Vertex deployment? No supported customer Azure OpenAI/Bedrock/Vertex setup or arbitrary OpenAI-compatible/custom endpoint configuration was retrieved. - custom_endpoint — Can an admin add any OpenAI-compatible endpoint? [unknown]: Can an admin add any OpenAI-compatible endpoint? No supported customer Azure OpenAI/Bedrock/Vertex setup or arbitrary OpenAI-compatible/custom endpoint configuration was retrieved. - byo_pricing — Does BYO change the licence price? [partial]: The paid subscription and direct-provider charges are separate in the vendor discussion. A discounted BYOK-only licence and exact customer provider rates are not verified. Sources: t3-chat:t3-chat:r2-paid-tier-controls. - byo_restrictions — Which features stop working with BYO? [unknown]: Current all-model BYOK compatibility is unknown: the OpenRouter-coverage request is marked Completed while the vendor says those model routes do not run through OpenRouter and rerouting requires future work. The old OpenAI GPT5 setup reply does not establish a complete current route matrix. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: t3-chat:t3-chat:r2-paid-tier-controls, t3-chat:t3-chat:r2-pro-context-byok ### MOD-04: Local and self-hosted models Judgment rule: Record support for models running in the customer's environment (vLLM, Ollama, TGI or similar) and which features work with them: chat, RAG, tools, embeddings. Distinguish a documented, supported route from a generic endpoint that happens to work. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial Vendor-operated open-model routes exist; customer-hosted LLM and embedding routes were not verified. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - vendor_hosted_open_models: Open-weight routes include GPT-OSS via Nebius in Finland, DeepSeek/Kimi via TensorX in Ireland and custom models via Verda in Finland; the active table also lists Melious routes. These are vendor-selected routes, with exact region absent for Melious. - Unverified self_hosted_chat, self_hosted_embeddings, feature_parity: Models documentation excludes adding customer model endpoints. No supported customer-hosted embedding route or parity matrix was found; Outpost stores a local index but does not establish local inference. - self_hosted_chat — Can chat use a customer-hosted model? [unknown]: Can chat use a customer-hosted model? Unknown. Models documentation excludes adding customer model endpoints. No supported customer-hosted embedding route or parity matrix was found; Outpost stores a local index but does not establish local inference. - self_hosted_embeddings — Can retrieval use customer-hosted embeddings? [unknown]: Can retrieval use customer-hosted embeddings? Unknown. Models documentation excludes adding customer model endpoints. No supported customer-hosted embedding route or parity matrix was found; Outpost stores a local index but does not establish local inference. - feature_parity — Which features fail with self-hosted models? [unknown]: Which features fail with self-hosted models? Unknown. Models documentation excludes adding customer model endpoints. No supported customer-hosted embedding route or parity matrix was found; Outpost stores a local index but does not establish local inference. - vendor_hosted_open_models — Does the vendor host open-weight models in EU/CH? [partial]: Open-weight routes include GPT-OSS via Nebius in Finland, DeepSeek/Kimi via TensorX in Ireland and custom models via Verda in Finland; the active table also lists Melious routes. These are vendor-selected routes, with exact region absent for Melious. Sources: meingpt:meingpt:lifecycle, meingpt:meingpt:subprocessors. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:lifecycle, meingpt:meingpt:subprocessors #### T3 Chat Verification status: No public information found (not-published) Support: not-published We found no public information on this. A missing finding is not evidence that the product lacks the capability. ### MOD-05: Model control and usage limits Judgment rule: Record admin allow-lists per group, default models, restrictions on which models may see which data, and per-user rate or fair-use limits. Test one restriction with two users in different groups. Record limits as enforced stops, warnings or undocumented fair use. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial Admins set model access and defaults; personal category limits have documented exceptions. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - limit_behaviour: Zero blocks the category; API budget exhaustion returns 403. Running/concurrent meeting recordings can exceed personal limits; scheduled tasks consume workspace credit outside personal categories. - Unverified data_bound_models: A separate rule binding particular knowledge sources to approved models was not documented; no two-user restriction test was run. - model_allowlist_by_group — Can admins limit models per group? [supported]: Per-team/person allow and deny rules are documented. Workspace admins and free models are exempt; denied chat models fall back, denied API models fail. Sources: meingpt:meingpt:config. - data_bound_models — Can admins restrict which models receive connector or knowledge content? [unknown]: Can admins restrict which models receive connector or knowledge content? Unknown. A separate rule binding particular knowledge sources to approved models was not documented; no two-user restriction test was run. - default_model — Can admins set the default model? [supported]: Workspace admins set the default model and chat standards. Sources: meingpt:meingpt:config. - user_limits — What per-user message or token limits apply, and are they published? [supported]: Six usage levels set monthly EUR limits for text, images, API, other services and meeting bots; no single universal message cap. Sources: meingpt:meingpt:limits. - limit_behaviour — What happens when a limit is reached? [partial]: Zero blocks the category; API budget exhaustion returns 403. Running/concurrent meeting recordings can exceed personal limits; scheduled tasks consume workspace credit outside personal categories. Sources: meingpt:meingpt:limits, meingpt:meingpt:api. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:config, meingpt:meingpt:limits, meingpt:meingpt:api #### T3 Chat Verification status: Agent cross-checked (agent-cross-checked) Support: partial Individuals have cost-based refill limits; enterprise group governance is unverified. Availability: generally-available Delivery: native Scope: surfaces=T3 Chat browser application; version=Public FAQ, legal pages and vendor replies retrieved 2026-10-21; no authenticated deployment observed; plan=Individual Pro USD8 cost-limit discussion and Premier account bucket reply; Free excluded from this mechanics finding; add_ons=Optional customer API keys; plan-specific quotas and checkout details not fully public; deployment_mode=not specified; region=US-hosted service with provider processing in US or other countries; no EU/CH-only option established; os=not specified. Limitations: - Documented findings only; unestablished atomic questions remain explicit gaps. - No complete Free/Pro/Premier entitlement matrix or authenticated model/feature menu was retrieved. - model_allowlist_by_group — Can admins limit models per group? [unknown]: Can admins limit models per group? No organization/group model allowlists, data-bound model policy, administrator default model or fixed per-user token/message quota was established. - data_bound_models — Can admins restrict which models receive connector or knowledge content? [unknown]: Can admins restrict which models receive connector or knowledge content? No organization/group model allowlists, data-bound model policy, administrator default model or fixed per-user token/message quota was established. - default_model — Can admins set the default model? [unknown]: Can admins set the default model? No organization/group model allowlists, data-bound model policy, administrator default model or fixed per-user token/message quota was established. - user_limits — What per-user message or token limits apply, and are they published? [supported]: Pro cost-limit replies and a Premier-specific vendor support reply tie paid usage to the FAQ Base/Overage mechanics: four-hour Base refill and monthly Overage refill. No fixed token/message allowance is published;20/200 example points are illustrative, not plan capacities. Sources: t3-chat:t3-chat:faq-usage, t3-chat:t3-chat:r2-paid-tier-controls, t3-chat:t3-chat:r2-premier-buckets. - limit_behaviour — What happens when a limit is reached? [supported]: For these documented paid-account bucket routes, expected request cost is reserved then settled after completion; when both balances are exhausted the FAQ says to wait for refill. No Free-tier behaviour is inferred. Sources: t3-chat:t3-chat:faq-usage, t3-chat:t3-chat:r2-premier-buckets, t3-chat:t3-chat:r2-paid-tier-controls. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: t3-chat:t3-chat:faq-usage, t3-chat:t3-chat:r2-paid-tier-controls, t3-chat:t3-chat:r2-premier-buckets ## Company knowledge ### KNOW-01: Files in chat Judgment rule: Record supported formats, size and count limits, scanned-PDF OCR, tables and spreadsheets, images inside documents, and whether a file is read in full or retrieved in chunks. Upload the shared fixture set (long German PDF, scanned contract, Excel with formulas, PowerPoint) and record failures. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial Chat uploads and direct sandbox processing have different document behavior. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - spreadsheets: Text extraction loses layout; the sandbox reads original Excel files using code and can preserve formulas. No fixture result is claimed. - full_vs_chunked: Direct attachments enter model context after extraction and are constrained by the context window; pools retrieve passages instead. Long-chat compression can drop earlier attachment context. - Unverified ocr: No scanned contract or structured spreadsheet fixture was uploaded; OCR behavior for direct scanned attachments was not bounded by a test. - formats — Which formats are accepted? [supported]: Text/PDF, images, Word, PowerPoint, Excel, CSV, TXT and URLs are listed; sandbox is needed to read archives and raw mail formats. Sources: meingpt:meingpt:flow. - limits — What size and count limits apply per file and chat? [supported]: 50 files per chat, 50 MB per non-image file; JPEG/PNG/WEBP/GIF are model-readable images. Provider PDF/page/image limits still apply. Sources: meingpt:meingpt:chat, meingpt:meingpt:models. - ocr — Are scanned PDFs read? [unknown]: Are scanned PDFs read? Unknown. No scanned contract or structured spreadsheet fixture was uploaded; OCR behavior for direct scanned attachments was not bounded by a test. - spreadsheets — Are Excel tables read with structure intact? [partial]: Text extraction loses layout; the sandbox reads original Excel files using code and can preserve formulas. No fixture result is claimed. Sources: meingpt:meingpt:chat, meingpt:meingpt:sandbox. - full_vs_chunked — Is the whole file used or only retrieved parts, and does the user see which? [partial]: Direct attachments enter model context after extraction and are constrained by the context window; pools retrieve passages instead. Long-chat compression can drop earlier attachment context. Sources: meingpt:meingpt:chat, meingpt:meingpt:models, meingpt:meingpt:pools. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:flow, meingpt:meingpt:chat, meingpt:meingpt:models, meingpt:meingpt:sandbox, meingpt:meingpt:pools #### T3 Chat Verification status: Agent cross-checked (agent-cross-checked) Support: partial Paid Pro context limits and own-key fallback are documented; plan-specific upload formats/caps remain unknown. Availability: generally-available Delivery: native Scope: surfaces=T3 Chat browser application; version=Public FAQ, legal pages and vendor replies retrieved 2026-10-21; no authenticated deployment observed; plan=Individual Pro USD8 context restrictions and optional BYOK fallback; exact upload entitlement by tier unverified; add_ons=Optional customer API keys; plan-specific quotas and checkout details not fully public; deployment_mode=not specified; region=US-hosted service with provider processing in US or other countries; no EU/CH-only option established; os=not specified. Limitations: - Documented findings only; unestablished atomic questions remain explicit gaps. - No complete Free/Pro/Premier entitlement matrix or authenticated model/feature menu was retrieved. - formats — Which formats are accepted? [unknown]: The privacy policy mentions PDF/images/text and the request is Completed, but no current individual-tier format entitlement was verified. The earlier sentence attributing browser/file-specific PDF upload bugs to a vendor reply is withdrawn because no precise reply locator was reproduced. - limits — What size and count limits apply per file and chat? [partial]: Vendor replies distinguish context restrictions on USD8 Pro from higher-tier allowances and own-key fallback. They do not publish a complete current per-model file size/count or context-cap matrix; no32K or user-suggested MB cap is adopted. Sources: t3-chat:t3-chat:r2-pro-context-byok. - ocr — Are scanned PDFs read? [unknown]: Are scanned PDFs read? No scanned-PDF OCR fixture, Excel structure/formula test, full-vs-chunked processing specification or current numerical upload matrix was retrieved. Community complaints and example sizes are not treated as product limits. - spreadsheets — Are Excel tables read with structure intact? [unknown]: Are Excel tables read with structure intact? No scanned-PDF OCR fixture, Excel structure/formula test, full-vs-chunked processing specification or current numerical upload matrix was retrieved. Community complaints and example sizes are not treated as product limits. - full_vs_chunked — Is the whole file used or only retrieved parts, and does the user see which? [unknown]: Is the whole file used or only retrieved parts, and does the user see which? No scanned-PDF OCR fixture, Excel structure/formula test, full-vs-chunked processing specification or current numerical upload matrix was retrieved. Community complaints and example sizes are not treated as product limits. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: t3-chat:t3-chat:r2-pro-context-byok ### KNOW-02: Knowledge bases and shared collections Judgment rule: Record curated document collections that persist across chats, who can create and share them, size limits, and how they attach to assistants. Distinguish personal, group and organisation-wide scope. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: supported Cloud pools persist across chats and attach to assistants; Outpost is a separate local index. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. - persistent_collections — Can users keep document collections across chats? [supported]: Persistent cloud data sources index SharePoint/OneDrive, Drive, Confluence, S3, WebDAV and web content. Sources: meingpt:meingpt:pools. - sharing_scope — Can collections be shared with users, groups or the whole organisation? [supported]: Admins share data sources with teams; workspace admins have broad visibility. Sources: meingpt:meingpt:pools. - size_limits — What are the collection limits? [supported]: Standard cloud raw-data budget is 300 GB across the workspace; indexing new content stops when full. Sources: meingpt:meingpt:pools. - assistant_binding — Can collections be bound to an assistant? [supported]: Pools bind to assistants or can be selected in a chat; projects themselves cannot bind pools. Sources: meingpt:meingpt:assistants, meingpt:meingpt:projects. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:pools, meingpt:meingpt:assistants, meingpt:meingpt:projects #### T3 Chat Verification status: No public information found (not-published) Support: not-published We found no public information on this. A missing finding is not evidence that the product lacks the capability. ### KNOW-03: Enterprise connectors Judgment rule: List each connector separately with plan, availability and setup owner: SharePoint, OneDrive, Outlook, Teams, Google Drive, Gmail, Confluence, Jira, Notion, Box, Salesforce, ServiceNow, file shares. Record whether the vendor or a third party (or the customer via MCP) provides it, and whether it needs admin consent. A connector listed on a roadmap or partner page is `announced`. An on-prem connector or gateway proves access to local data, not a self-hosted platform (DATA-02). #### meinGPT Verification status: No public information found (not-published) Support: not-published We found no public information on this. A missing finding is not evidence that the product lacks the capability. #### T3 Chat Verification status: No public information found (not-published) Support: not-published We found no public information on this. A missing finding is not evidence that the product lacks the capability. ### KNOW-04: Permission-aware retrieval Judgment rule: Using two test users, verify that a restricted document in the fixture tenant is never cited or summarised for the user without access, including after a permission change. Record whether permissions are enforced at query time by the source system, mirrored into an index, or not enforced. Record the propagation delay of a revocation. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial Live OAuth preserves source access; indexed permissions require a separate, default-off setting. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - source_permissions_respected: Live M365 tools use each user account. Cloud SharePoint/Drive pools enforce imported source permissions only when request-enabled and switched on; other source types do not inherit them. No two-user test was run. - Unverified revocation_delay, sensitivity_labels: No restricted-document/revocation fixture or Purview-label check was run; revocation timing and sensitivity-label behavior remain unknown. - source_permissions_respected — Are source permissions enforced per user? [partial]: Live M365 tools use each user account. Cloud SharePoint/Drive pools enforce imported source permissions only when request-enabled and switched on; other source types do not inherit them. No two-user test was run. Sources: meingpt:meingpt:m365, meingpt:meingpt:pool-controls. - enforcement_point — Does the source system, a mirrored index or nothing enforce them? [supported]: Live tools execute under user OAuth. Indexed SharePoint/Drive permissions are mirrored and fail closed when stale; source-permission mode is off by default. Sources: meingpt:meingpt:pool-controls. - revocation_delay — How quickly does a removed permission take effect? [unknown]: How quickly does a removed permission take effect? Unknown. No restricted-document/revocation fixture or Purview-label check was run; revocation timing and sensitivity-label behavior remain unknown. - admin_scope_controls — Can admins exclude sites, folders or labels? [supported]: Admins select imported sources/folders and control team sharing. Sources: meingpt:meingpt:pools. - sensitivity_labels — Are Microsoft Purview or similar labels respected? [unknown]: Are Microsoft Purview or similar labels respected? Unknown. No restricted-document/revocation fixture or Purview-label check was run; revocation timing and sensitivity-label behavior remain unknown. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:m365, meingpt:meingpt:pool-controls, meingpt:meingpt:pools #### T3 Chat Verification status: No public information found (not-published) Support: not-published We found no public information on this. A missing finding is not evidence that the product lacks the capability. ### KNOW-05: Freshness and sync model Judgment rule: Record live (federated) search versus a copied index, sync interval, deletion propagation, and where the index is stored. Edit and delete one fixture document and record when answers change. A copied index adds storage location to DATA-01. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial Live connectors query the source; cloud pools copy an index to Germany. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - Unverified deletion_propagation: No edit/delete fixture was run and an end-to-end deletion-propagation bound was not verified. - live_or_indexed — Is content searched live or copied into an index? [supported]: Native connectors fetch live; cloud data pools index imported documents. Sources: meingpt:meingpt:pools. - sync_interval — How often does an index refresh? [supported]: Cloud import runs every 15 minutes. Sources: meingpt:meingpt:pools. - deletion_propagation — How quickly does a deleted document disappear? [unknown]: How quickly does a deleted document disappear? Unknown. No edit/delete fixture was run and an end-to-end deletion-propagation bound was not verified. - index_location — Where is indexed content stored? [supported]: Cloud pools run at Hetzner in Germany; local Outpost pools stay in the customer network. Sources: meingpt:meingpt:pools. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:pools #### T3 Chat Verification status: No public information found (not-published) Support: not-published We found no public information on this. A missing finding is not evidence that the product lacks the capability. ### KNOW-06: Grounding and retrieval quality Judgment rule: Run the shared question set over the fixture corpus: factual lookups, cross-document questions, German document with an English question, a question with no answer in the corpus. Record citation presence, link accuracy, correct answers and refusals to invent. Disclose model, settings, attempt count and failures. Single runs support a case study, not a ranking. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial Source display is documented; grounding quality remains untested. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - no_answer_behaviour: The guide says source search reports no matching result; actual refusals were not tested. - Unverified citation_accuracy, cross_language, observed_accuracy: The shared multilingual corpus and question set were not run, so link correctness, cross-language retrieval and accuracy counts are unknown. - citations — Are answers linked to specific sources? [supported]: Cloud source answers always display sources and this cannot be disabled. Sources: meingpt:meingpt:pool-controls. - citation_accuracy — Do the cited passages support the answer? [unknown]: Do the cited passages support the answer? Unknown. The shared multilingual corpus and question set were not run, so link correctness, cross-language retrieval and accuracy counts are unknown. - cross_language — Does a question in one language find documents in another? [unknown]: Does a question in one language find documents in another? Unknown. The shared multilingual corpus and question set were not run, so link correctness, cross-language retrieval and accuracy counts are unknown. - no_answer_behaviour — Does it say when the corpus has no answer? [partial]: The guide says source search reports no matching result; actual refusals were not tested. Sources: meingpt:meingpt:pools. - observed_accuracy — How many of the question set were answered correctly? [unknown]: How many of the question set were answered correctly? Unknown. The shared multilingual corpus and question set were not run, so link correctness, cross-language retrieval and accuracy counts are unknown. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:pool-controls, meingpt:meingpt:pools #### T3 Chat Verification status: No public information found (not-published) Support: not-published We found no public information on this. A missing finding is not evidence that the product lacks the capability. ### KNOW-07: Memory and personal context Judgment rule: Record personal custom instructions, cross-chat memory, project or workspace context, and user controls to view, edit, delete or disable memory. Record whether admins can disable memory organisation-wide. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial Personal instructions and user-specific assistant memory are documented. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - user_memory_controls: Users view and delete one/all memories; direct editing is unavailable in settings, but an assistant can update entries. - custom_instructions — Can users set persistent instructions? [supported]: Preferences store a persistent system instruction applied with workspace and assistant context. Sources: meingpt:meingpt:preferences. - cross_chat_memory — Does the product remember facts across chats? [supported]: Enabled assistants share a personal user memory store across chats, not a team store. Sources: meingpt:meingpt:memory. - user_memory_controls — Can users view, edit and delete memories? [partial]: Users view and delete one/all memories; direct editing is unavailable in settings, but an assistant can update entries. Sources: meingpt:meingpt:memory. - admin_memory_controls — Can admins turn memory off? [supported]: Admins can disable or remove Memory from the permitted tools. Sources: meingpt:meingpt:memory. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:preferences, meingpt:meingpt:memory #### T3 Chat Verification status: No public information found (not-published) Support: not-published We found no public information on this. A missing finding is not evidence that the product lacks the capability. ## Assistants and agents ### AGT-01: Custom assistants Judgment rule: Record creation of assistants with instructions, model, knowledge and tools; who can create; sharing scope; versioning; and an internal catalogue or directory. Build one HR policy assistant on the fixture and share it with one group. Prompt templates and a prompt library are recorded here. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: supported Assistants combine model, instructions, tools and knowledge, with governed creation and sharing. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. - create_assistants — Can users create assistants with instructions, knowledge and tools? [supported]: The editor supports instructions, model, files, pools, connectors and built-in tools. Sources: meingpt:meingpt:assistants. - creator_permissions — Can admins limit who creates them? [supported]: Request-enabled permissions restrict assistant creation/sharing/publication by team or person; admins are exempt. Sources: meingpt:meingpt:permissions. - sharing_scope — Can assistants be shared with users, groups or everyone? [supported]: Team and direct-person assistant sharing are documented; workspace-wide publication is admin-only by default. Sources: meingpt:meingpt:permissions, meingpt:meingpt:roles. - catalogue — Is there an internal assistant directory? [supported]: The navigation lists personal assistants and shared resources. Sources: meingpt:meingpt:assistants. - prompt_library — Are shared prompt templates available? [supported]: Prompt folders can be personal, workspace or department scoped. Sources: meingpt:meingpt:prompts. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:assistants, meingpt:meingpt:permissions, meingpt:meingpt:roles, meingpt:meingpt:prompts #### T3 Chat Verification status: No public information found (not-published) Support: not-published We found no public information on this. A missing finding is not evidence that the product lacks the capability. ### AGT-02: Tools, actions and MCP Judgment rule: Record built-in tools, custom actions (OpenAPI), MCP client support (local and remote servers), authentication model per tool (shared service account or per-user OAuth), and admin approval of tools. Verify one external-tool call and its access boundary. #### meinGPT Verification status: No public information found (not-published) Support: not-published We found no public information on this. A missing finding is not evidence that the product lacks the capability. #### T3 Chat Verification status: Agent cross-checked (agent-cross-checked) Support: partial Paid Pro search is discussed by the vendor; image-tier entitlement, custom actions and MCP remain unverified. Availability: generally-available Delivery: native Scope: surfaces=T3 Chat browser application; version=Public FAQ, legal pages and vendor replies retrieved 2026-10-21; no authenticated deployment observed; plan=Individual Pro USD8 for evidenced search use; image and MCP plan entitlement unknown; add_ons=Optional customer API keys; plan-specific quotas and checkout details not fully public; deployment_mode=not specified; region=US-hosted service with provider processing in US or other countries; no EU/CH-only option established; os=not specified. Limitations: - Documented findings only; unestablished atomic questions remain explicit gaps. - No complete Free/Pro/Premier entitlement matrix or authenticated model/feature menu was retrieved. - builtin_tools — Which tools are built in? [partial]: The Pro/Premier pricing discussion explicitly describes frequent search and search-token overhead in the USD8 paid context. An OpenAI image integration has separate support evidence, but its plan entitlement is not verified and is not included as a Pro tool promise. Sources: t3-chat:t3-chat:r2-paid-tier-controls. - custom_actions — Can admins add API actions? [unknown]: Can admins add API actions? The MCP request is Closed but has no retrieved released setup guide. Remote MCP configuration, custom API actions, user/shared tool credentials and administrator/user approval policies were not established. - mcp_client — Can it use remote MCP servers, and with what admin control? [unknown]: Can it use remote MCP servers, and with what admin control? The MCP request is Closed but has no retrieved released setup guide. Remote MCP configuration, custom API actions, user/shared tool credentials and administrator/user approval policies were not established. - tool_auth — Do tools act as the user or as a shared account? [unknown]: Do tools act as the user or as a shared account? The MCP request is Closed but has no retrieved released setup guide. Remote MCP configuration, custom API actions, user/shared tool credentials and administrator/user approval policies were not established. - tool_approval — Can admins approve or block tools, and can users confirm actions before they run? [unknown]: Can admins approve or block tools, and can users confirm actions before they run? The MCP request is Closed but has no retrieved released setup guide. Remote MCP configuration, custom API actions, user/shared tool credentials and administrator/user approval policies were not established. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: t3-chat:t3-chat:r2-paid-tier-controls ### AGT-03: Web search and deep research Judgment rule: Record web search on/off control per organisation and group, search provider, citation of web sources, deep or multi-step research, and whether queries leave the residency boundary. Admins of regulated companies often need web search off. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: supported Linkup powers cited web search and Deep Research; admins can restrict app access. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. - web_search — Is web search available? [supported]: Search can be switched on in chat, is always on in MeinGPT Web and runs in the Recherche app. Sources: meingpt:meingpt:web. - admin_toggle — Can admins disable it per organisation or group? [supported]: Request-enabled per-team/person app controls include Web search. Sources: meingpt:meingpt:permissions. - search_provider — Which search provider handles queries, and where? [supported]: Linkup processes search; the register lists Azure France. Prompt terms can enter generated search queries. Sources: meingpt:meingpt:web, meingpt:meingpt:subprocessors. - deep_research — Is multi-step research with a report available? [supported]: The research app has a user-controlled multi-step Deep Research mode. Sources: meingpt:meingpt:web. - web_citations — Are web sources cited? [supported]: Results include source previews and page references. Sources: meingpt:meingpt:web. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:web, meingpt:meingpt:permissions, meingpt:meingpt:subprocessors #### T3 Chat Verification status: Agent cross-checked (agent-cross-checked) Support: partial Search is user-selectable, with a documented one-to-five search limit; a separate deep-research product is unverified. Availability: generally-available Delivery: native Scope: surfaces=T3 Chat browser application; version=Public FAQ, legal pages and vendor replies retrieved 2026-10-21; no authenticated deployment observed; plan=Individual Pro USD8 paid-search discussion; no Free parity asserted; add_ons=Optional customer API keys; plan-specific quotas and checkout details not fully public; deployment_mode=not specified; region=US-hosted service with provider processing in US or other countries; no EU/CH-only option established; os=not specified. Limitations: - Documented findings only; unestablished atomic questions remain explicit gaps. - No complete Free/Pro/Premier entitlement matrix or authenticated model/feature menu was retrieved. - web_search — Is web search available? [partial]: Staff’s paid USD8 discussion confirms search consumes API/query and context costs. The FAQ exposes search on/off and1..5 URL limits, but a complete current tier-specific limit/feature matrix was not verified. Sources: t3-chat:t3-chat:r2-paid-tier-controls, t3-chat:t3-chat:faq-url. - admin_toggle — Can admins disable it per organisation or group? [unknown]: Can admins disable it per organisation or group? No organization/group switch, named search provider and query country, dedicated multi-step research/report mode or precise web-citation contract was established. The FAQ says search improves recency without guaranteeing correctness. - search_provider — Which search provider handles queries, and where? [unknown]: Which search provider handles queries, and where? No organization/group switch, named search provider and query country, dedicated multi-step research/report mode or precise web-citation contract was established. The FAQ says search improves recency without guaranteeing correctness. - deep_research — Is multi-step research with a report available? [unknown]: Is multi-step research with a report available? No organization/group switch, named search provider and query country, dedicated multi-step research/report mode or precise web-citation contract was established. The FAQ says search improves recency without guaranteeing correctness. - web_citations — Are web sources cited? [unknown]: Are web sources cited? No organization/group switch, named search provider and query country, dedicated multi-step research/report mode or precise web-citation contract was established. The FAQ says search improves recency without guaranteeing correctness. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: t3-chat:t3-chat:r2-paid-tier-controls, t3-chat:t3-chat:faq-url ### AGT-04: Workflows and automation Judgment rule: Record no-code or low-code workflows, scheduled prompts, triggers from email, forms or business systems, and multi-step chains. Name where they run and who can build them. A shared prompt template is not a workflow. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: supported Multi-step workflows are documented; scheduled/event automations are request-enabled beta. Availability: beta Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - Workflow editor is documented separately from request-enabled beta scheduled tasks; beta is not universal plan availability. - workflow_builder — Is there a workflow or chain builder? [supported]: A project editor chains prompts with variables and per-step models, or Buddy drafts the workflow. Sources: meingpt:meingpt:workflows. - scheduled_tasks — Can prompts or assistants run on a schedule? [supported]: Support enables beta tasks with daily, weekly, monthly or selected calendar schedules. Sources: meingpt:meingpt:scheduled. - event_triggers — Can external events start a workflow? [supported]: Beta tasks accept tokenized webhooks and task-specific email addresses. Sources: meingpt:meingpt:scheduled. - builder_audience — Is building aimed at business users, IT or developers? [supported]: Members can build with natural-language Buddy help or manual steps, subject to admin rights. Sources: meingpt:meingpt:workflows. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:workflows, meingpt:meingpt:scheduled #### T3 Chat Verification status: No public information found (not-published) Support: not-published We found no public information on this. A missing finding is not evidence that the product lacks the capability. ### AGT-05: Agentic task execution Judgment rule: Record an agent mode that plans and carries out multi-step tasks with tools (browser, files, connectors), with visible steps, user confirmation of side effects and stop control. Run one fixture task (gather three documents, draft a summary, propose an email). Record interventions and what ran. A final claim of success is not proof. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial Assistants can execute successive tool calls, show progress and stop; per-action confirmation is absent. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - side_effect_confirmation: No built-in method-specific permission popup exists at runtime; admin-approved write/dangerous methods can run without an additional dialog. - agent_mode — Is there an agent mode for multi-step tasks? [supported]: Assistant tools run multiple sequential steps within a response and stop at a bounded tool-call budget. Sources: meingpt:meingpt:assistants. - visible_steps — Are steps and tool calls shown? [supported]: Tool/reasoning steps and progress are shown in chat. Sources: meingpt:meingpt:assistants, meingpt:meingpt:chat. - side_effect_confirmation — Must the user confirm actions with side effects? [unsupported]: No built-in method-specific permission popup exists at runtime; admin-approved write/dangerous methods can run without an additional dialog. Sources: meingpt:meingpt:tools. - stop_control — Can the user stop a running task? [supported]: The input button changes to Abbrechen and stops response generation while preserving text. Sources: meingpt:meingpt:chat. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:assistants, meingpt:meingpt:chat, meingpt:meingpt:tools #### T3 Chat Verification status: No public information found (not-published) Support: not-published We found no public information on this. A missing finding is not evidence that the product lacks the capability. ### AGT-06: Programmable access Judgment rule: Record a public API for chat, assistants and knowledge, SDKs, API authentication, rate limits and whether API use counts against the same governance (audit, budgets, model restrictions). Being an MCP client is different from exposing the product over MCP. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial Enterprise-enabled APIs cover chat, assistants and workflows with model and budget enforcement. Availability: generally-available Delivery: native Scope: surfaces=meinGPT Enterprise API; version=Public documentation checked 2026-10-21; plan=Enterprise API, explicitly enabled; no public add-on price; add_ons=Enterprise API; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - api_governance: Model permission failures return 403; API has a separate personal budget and shared token-rate ceilings; compliance-content logging coverage was not fully established. - Unverified knowledge_api, mcp_server: A knowledge-ingestion API and product MCP-server endpoint were not established in the API guide. - chat_api — Is there an API for chat and assistants? [supported]: Bearer-key chat completions, assistants, workflow runs, embeddings and transcription are documented; personal keys require Enterprise enablement. Sources: meingpt:meingpt:api, meingpt:meingpt:outlook. - knowledge_api — Can documents be loaded through an API? [unknown]: Can documents be loaded through an API? Unknown. A knowledge-ingestion API and product MCP-server endpoint were not established in the API guide. - api_governance — Do audit, budgets and model rules apply to API use? [partial]: Model permission failures return 403; API has a separate personal budget and shared token-rate ceilings; compliance-content logging coverage was not fully established. Sources: meingpt:meingpt:api, meingpt:meingpt:config, meingpt:meingpt:audit. - mcp_server — Does the product expose itself to other agents over MCP? [unknown]: Does the product expose itself to other agents over MCP? Unknown. A knowledge-ingestion API and product MCP-server endpoint were not established in the API guide. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:api, meingpt:meingpt:outlook, meingpt:meingpt:config, meingpt:meingpt:audit #### T3 Chat Verification status: Agent cross-checked (agent-cross-checked) Support: partial No official programmable enterprise interface was established; subscription-backed external-agent API usage was rejected. Availability: generally-available Delivery: native Scope: surfaces=T3 Chat browser application; version=Public FAQ, legal pages and vendor replies retrieved 2026-10-21; no authenticated deployment observed; plan=Individual Free, Pro and Premier evidence; no organizational plan or entitlement established; add_ons=Optional customer API keys; plan-specific quotas and checkout details not fully public; deployment_mode=not specified; region=US-hosted service with provider processing in US or other countries; no EU/CH-only option established; os=not specified. Limitations: - Documented findings only; unestablished atomic questions remain explicit gaps. - chat_api — Is there an API for chat and assistants? [partial]: An official feedback request asks to apply subscription quota to external coding agents/API; staff closes it because the economics do not work. The FAQ URL launches a browser chat and is not a documented chat API. Sources: t3-chat:t3-chat:api-request, t3-chat:t3-chat:faq-url. - knowledge_api — Can documents be loaded through an API? [unknown]: Can documents be loaded through an API? No supported general chat/assistant API, document-ingestion API, API governance or MCP server exposure was retrieved. Third-party cookie/API clients and T3 Code repositories are not official T3 Chat API evidence. - api_governance — Do audit, budgets and model rules apply to API use? [unknown]: Do audit, budgets and model rules apply to API use? No supported general chat/assistant API, document-ingestion API, API governance or MCP server exposure was retrieved. Third-party cookie/API clients and T3 Code repositories are not official T3 Chat API evidence. - mcp_server — Does the product expose itself to other agents over MCP? [unknown]: Does the product expose itself to other agents over MCP? No supported general chat/assistant API, document-ingestion API, API governance or MCP server exposure was retrieved. Third-party cookie/API clients and T3 Code repositories are not official T3 Chat API evidence. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: t3-chat:t3-chat:api-request, t3-chat:t3-chat:faq-url ## Daily work ### WORK-01: Clients and access points Judgment rule: List web, desktop (Windows, macOS), iOS, Android, mobile browser and browser extension separately. For each, record login with company SSO, feature gaps against web, and MDM/Intune deployability. An installable app is different from a mobile website. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial The platform is browser-based; Outlook has a separate centrally deployed add-in. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace, meinGPT Outlook add-in, meinGPT Enterprise API; version=Public documentation checked 2026-10-21; plan=Business/Enterprise web workspace; Outlook add-in requires Enterprise API enablement and personal API key; add_ons=Enterprise API enabled for Outlook add-in; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - managed_deployment: The Outlook add-in can be assigned to users or groups in Microsoft 365 integrated apps; general desktop/mobile MDM configuration was not found. - Unverified desktop_apps, mobile_apps, browser_extension, client_feature_gaps: No first-party native Windows/macOS/iOS/Android workspace client or browser-extension support matrix was located in the user guide, client setup or official site search; mobile parity was not tested. - web — Is a full web client available? [supported]: The terms define a browser-accessible workspace. Sources: meingpt:meingpt:terms. - desktop_apps — Windows and macOS apps? [unknown]: Windows and macOS apps? Unknown. No first-party native Windows/macOS/iOS/Android workspace client or browser-extension support matrix was located in the user guide, client setup or official site search; mobile parity was not tested. - mobile_apps — iOS and Android apps with SSO? [unknown]: iOS and Android apps with SSO? Unknown. No first-party native Windows/macOS/iOS/Android workspace client or browser-extension support matrix was located in the user guide, client setup or official site search; mobile parity was not tested. - browser_extension — Is there a browser extension or sidebar? [unknown]: Is there a browser extension or sidebar? Unknown. No first-party native Windows/macOS/iOS/Android workspace client or browser-extension support matrix was located in the user guide, client setup or official site search; mobile parity was not tested. - managed_deployment — Can IT deploy and configure clients through MDM? [partial]: The Enterprise API-enabled Outlook add-in can be assigned to users/groups through Microsoft365 integrated apps. It supports Outlook web, Windows classic/new and Mac; each user needs a personal meinGPT API key. General desktop/mobile MDM configuration was not established. Sources: meingpt:meingpt:outlook. - client_feature_gaps — Which features are missing on mobile? [unknown]: Which features are missing on mobile? Unknown. No first-party native Windows/macOS/iOS/Android workspace client or browser-extension support matrix was located in the user guide, client setup or official site search; mobile parity was not tested. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:terms, meingpt:meingpt:outlook #### T3 Chat Verification status: Agent cross-checked (agent-cross-checked) Support: partial A browser client and iOS TestFlight beta are documented; desktop, Android release and managed clients remain unverified. Availability: generally-available Delivery: native Scope: surfaces=T3 Chat browser application, T3 Chat iOS TestFlight beta; version=Public FAQ, legal pages and vendor replies retrieved 2026-10-21; no authenticated deployment observed; plan=Browser account plan; iOS TestFlight beta; enterprise SSO unverified; add_ons=Optional customer API keys; plan-specific quotas and checkout details not fully public; deployment_mode=not specified; region=US-hosted service with provider processing in US or other countries; no EU/CH-only option established; os=iOS. Limitations: - Documented findings only; unestablished atomic questions remain explicit gaps. - web — Is a full web client available? [supported]: T3 Chat is delivered as a browser service; the FAQ documents launching new conversations through public URLs. Sources: t3-chat:t3-chat:faq-url. - desktop_apps — Windows and macOS apps? [unknown]: Windows and macOS apps? No native Windows/macOS distribution, supported browser extension/sidebar, MDM configuration or complete mobile feature-gap matrix was established. Community extension links are not vendor clients. - mobile_apps — iOS and Android apps with SSO? [partial]: The vendor’s September10,2026 announcement releases iOS via TestFlight explicitly as beta; Android beta is coming soon. Neither corporate SSO nor a production app-store release was verified. Sources: t3-chat:t3-chat:mobile. - browser_extension — Is there a browser extension or sidebar? [unknown]: Is there a browser extension or sidebar? No native Windows/macOS distribution, supported browser extension/sidebar, MDM configuration or complete mobile feature-gap matrix was established. Community extension links are not vendor clients. - managed_deployment — Can IT deploy and configure clients through MDM? [unknown]: Can IT deploy and configure clients through MDM? No native Windows/macOS distribution, supported browser extension/sidebar, MDM configuration or complete mobile feature-gap matrix was established. Community extension links are not vendor clients. - client_feature_gaps — Which features are missing on mobile? [unknown]: Which features are missing on mobile? No native Windows/macOS distribution, supported browser extension/sidebar, MDM configuration or complete mobile feature-gap matrix was established. Community extension links are not vendor clients. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: t3-chat:t3-chat:faq-url, t3-chat:t3-chat:mobile ### WORK-02: Office suite and Teams integration Judgment rule: Record presence inside Word, Excel, PowerPoint, Outlook and Teams (and Google Docs, Gmail), with what the assistant can do in each: read the open document, write into it, act on mail. Record extra licences and admin consent needed. A connector that searches Office files is KNOW-03, not WORK-02. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial A documented Outlook add-in operates in the received-message view and needs Enterprise API access. Availability: generally-available Delivery: native Scope: surfaces=meinGPT Outlook add-in; version=Public documentation checked 2026-10-21; plan=Enterprise API enabled workspace; add_ons=Enterprise API; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - licence_prerequisites: Personal meinGPT API key and Enterprise API enablement are required; tenant admins can centrally deploy the add-in. A separate Microsoft licence price is not stated. - Unverified word_excel_powerpoint, teams, google_workspace_apps: Word/Excel/PowerPoint/Teams or Google-editor in-app extensions were not established; connectors that read their content are recorded in KNOW-03. - word_excel_powerpoint — Does it work inside Word, Excel and PowerPoint? [unknown]: Does it work inside Word, Excel and PowerPoint? Unknown. Word/Excel/PowerPoint/Teams or Google-editor in-app extensions were not established; connectors that read their content are recorded in KNOW-03. - outlook — Does it work inside Outlook? [supported]: Outlook Web, Windows classic/new and Mac support summarizing the selected received email and drafting replies; compose view is excluded. Sources: meingpt:meingpt:outlook. - teams — Is there a Teams app or bot? [unknown]: Is there a Teams app or bot? Unknown. Word/Excel/PowerPoint/Teams or Google-editor in-app extensions were not established; connectors that read their content are recorded in KNOW-03. - google_workspace_apps — Does it work inside Docs, Sheets or Gmail? [unknown]: Does it work inside Docs, Sheets or Gmail? Unknown. Word/Excel/PowerPoint/Teams or Google-editor in-app extensions were not established; connectors that read their content are recorded in KNOW-03. - licence_prerequisites — Which licences or consents are needed? [partial]: Personal meinGPT API key and Enterprise API enablement are required; tenant admins can centrally deploy the add-in. A separate Microsoft licence price is not stated. Sources: meingpt:meingpt:outlook. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:outlook #### T3 Chat Verification status: No public information found (not-published) Support: not-published We found no public information on this. A missing finding is not evidence that the product lacks the capability. ### WORK-03: Translation and writing Judgment rule: Record document translation preserving layout, the translation engine (LLM or dedicated such as DeepL), glossaries, formality control, supported languages including Swiss German conventions (no ß), and rewriting/proofreading tools. Translate the shared German contract into French and Italian and record layout and glossary results. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial Dedicated text/document translation uses EU Google/DeepL routes and managed glossaries. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - document_translation: DeepL document translation preserves layout for DOC/DOCX, PPT/PPTX, XLS/XLSX, PDF, HTML, TXT, SRT and XLIFF. Layout quality was not fixture-tested. - writing_tools: Quick Actions rewrite answers, including more formal and improve options. - Unverified formality_and_variants: Formality settings within the translator and guaranteed de-CH spelling were not established; no DE-to-FR/IT contract fixture was translated. - document_translation — Can whole documents be translated with formatting kept? [partial]: DeepL document translation preserves layout for DOC/DOCX, PPT/PPTX, XLS/XLSX, PDF, HTML, TXT, SRT and XLIFF. Layout quality was not fixture-tested. Sources: meingpt:meingpt:translation. - translation_engine — Which engine translates? [supported]: Default Google Cloud Translate for text and DeepL for documents; own DeepL key switches both to DeepL. Sources: meingpt:meingpt:translation. - glossaries — Are company glossaries supported? [supported]: Workspace admins create per-language-pair glossaries and import CSV; exact word forms must match. Sources: meingpt:meingpt:translation. - formality_and_variants — Can formality and regional variants (de-CH) be set? [unknown]: Can formality and regional variants (de-CH) be set? Unknown. Formality settings within the translator and guaranteed de-CH spelling were not established; no DE-to-FR/IT contract fixture was translated. - writing_tools — Are dedicated rewrite and proofreading tools available? [partial]: Quick Actions rewrite answers, including more formal and improve options. Sources: meingpt:meingpt:chat. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:translation, meingpt:meingpt:chat #### T3 Chat Verification status: No public information found (not-published) Support: not-published We found no public information on this. A missing finding is not evidence that the product lacks the capability. ### WORK-04: Image generation and editing Judgment rule: Record available image models, editing (inpainting, reference images), admin control and per-image cost visibility, and licence or usage terms for generated images. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial Twelve named image variants include reference editing with administrator model controls. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - image_editing: Reference image generation supports model-dependent uploads; app lacks post-generation crop/color editing. - usage_terms: Terms warn third-party output may carry rights; no blanket unrestricted-use warranty is made. C2PA metadata labels generated output. - image_generation — Can users generate images, with which models? [supported]: FLUX.2 Klein/Pro/Flex/Max, FLUX 1.1 Pro, FLUX.1 Kontext Max; GPT Image 1/Mini/1.5/2; Gemini 2.5 Flash Image and 3 Pro Image. Sources: meingpt:meingpt:images. - image_editing — Can users edit uploaded images? [partial]: Reference image generation supports model-dependent uploads; app lacks post-generation crop/color editing. Sources: meingpt:meingpt:images. - admin_control — Can admins restrict image generation? [supported]: Admins enable models and default; request-enabled permissions can restrict Image app use by teams or persons. Sources: meingpt:meingpt:image-admin, meingpt:meingpt:permissions. - usage_terms — What usage rights apply to output? [partial]: Terms warn third-party output may carry rights; no blanket unrestricted-use warranty is made. C2PA metadata labels generated output. Sources: meingpt:meingpt:terms, meingpt:meingpt:images. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:images, meingpt:meingpt:image-admin, meingpt:meingpt:permissions, meingpt:meingpt:terms #### T3 Chat Verification status: No public information found (not-published) Support: not-published We found no public information on this. A missing finding is not evidence that the product lacks the capability. ### WORK-05: Voice and meetings Judgment rule: Record dictation, read-aloud, live voice conversation, meeting transcription and summarisation, supported languages including Swiss German speech, and where audio is processed. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial Dictation and meeting transcription are documented; speech quality was not tested. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - Unverified read_aloud, voice_conversation, swiss_german_speech: No first-party complete read-aloud/live-voice plan and language matrix or Swiss German recognition fixture was established. - dictation — Can users dictate prompts? [supported]: A microphone transcribes prompts into the input field. Sources: meingpt:meingpt:chat. - read_aloud — Can answers be read aloud? [unknown]: Can answers be read aloud? Unknown. No first-party complete read-aloud/live-voice plan and language matrix or Swiss German recognition fixture was established. - voice_conversation — Is a live voice mode available? [unknown]: Is a live voice mode available? Unknown. No first-party complete read-aloud/live-voice plan and language matrix or Swiss German recognition fixture was established. - meeting_summaries — Are meetings transcribed and summarised? [supported]: Audio/video/transcript uploads and meeting-bot notes are documented; storage Hetzner Germany, Gladia transcription France, Vertex EU summaries. Sources: meingpt:meingpt:meeting. - swiss_german_speech — Is Swiss German speech recognised? [unknown]: Is Swiss German speech recognised? Unknown. No first-party complete read-aloud/live-voice plan and language matrix or Swiss German recognition fixture was established. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:chat, meingpt:meingpt:meeting #### T3 Chat Verification status: No public information found (not-published) Support: not-published We found no public information on this. A missing finding is not evidence that the product lacks the capability. ### WORK-06: Data analysis and document output Judgment rule: Record code execution for data analysis, charts, and generation of downloadable Word, Excel, PowerPoint and PDF files. Analyse the fixture Excel and request a PowerPoint summary. Record where code runs and whether files are editable. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial An isolated EU sandbox executes analysis code and generates downloadable Office files. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - canvas_editing: Downloaded artifacts and an Excel editor exist; editor changes stay local until downloaded and the chat card keeps the original. - data_analysis — Can it analyse spreadsheets with executed code? [supported]: Python/JavaScript sandbox reads original spreadsheets with pandas/openpyxl and executes code. Sources: meingpt:meingpt:sandbox. - charts — Can it produce charts? [supported]: Executed matplotlib charts can be returned as PNG. Sources: meingpt:meingpt:sandbox. - office_file_output — Can it create Word, Excel or PowerPoint files? [supported]: Sandbox libraries support DOCX/XLSX/PPTX; workflows also generate Word/PDF/Excel/CSV/PowerPoint/JSON. Sources: meingpt:meingpt:sandbox, meingpt:meingpt:workflows. - canvas_editing — Is there a document or canvas editor for iterating on output? [partial]: Downloaded artifacts and an Excel editor exist; editor changes stay local until downloaded and the chat card keeps the original. Sources: meingpt:meingpt:sandbox. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:sandbox, meingpt:meingpt:workflows #### T3 Chat Verification status: No public information found (not-published) Support: not-published We found no public information on this. A missing finding is not evidence that the product lacks the capability. ### WORK-07: Sharing and team spaces Judgment rule: Record chat sharing (link, user, group), shared projects or team spaces, channels or group chat with AI, and commenting. Test sharing a chat with a colleague in another group and record what the recipient sees, including knowledge they lack access to. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial Chat links and shared projects exist; projects preserve private member chat histories. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - group_chat: Each project member has a separate private chat list; shared project chats are not a simultaneous group conversation. - Unverified share_permission_boundary: No cross-group chat-sharing test was run; recipient access to source attachments and restricted knowledge was not verified. - Project chat privacy does not prove absence of every group-chat surface. - share_chats — Can chats be shared inside the organisation? [supported]: A share toggle creates a link displaying the shared conversation. Sources: meingpt:meingpt:chat. - team_spaces — Are shared projects or workspaces available? [supported]: Projects share instructions and workflows with Owner/Editor/Member roles. Sources: meingpt:meingpt:projects. - group_chat — Can several people chat together with the AI? [partial]: Within Projects,each member has a separate private chat list. Simultaneous multi-user group conversation elsewhere was not established. Sources: meingpt:meingpt:projects. - share_permission_boundary — Does sharing respect the recipient's access to sources? [unknown]: Does sharing respect the recipient's access to sources? Unknown. No cross-group chat-sharing test was run; recipient access to source attachments and restricted knowledge was not verified. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:chat, meingpt:meingpt:projects #### T3 Chat Verification status: No public information found (not-published) Support: not-published We found no public information on this. A missing finding is not evidence that the product lacks the capability. ### WORK-08: Interface languages and accessibility Judgment rule: Record UI languages, focusing on German, French, Italian and English, with translation quality observed (complete, partial, machine-translated). Record response language handling and WCAG conformance statements. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial Platform-language choice exists, but the full UI and accessibility matrix remains unverified. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - ui_languages: Preferences provide a platform-language selector distinct from response language; a complete named language list is not published on this page. - Unverified dach_language_quality, accessibility_statement, admin_ui_languages: No authenticated DE/FR/IT/EN UI inspection or WCAG conformance statement was located in the user guide or security pages; admin language completeness is untested. - ui_languages — Which UI languages are offered? [partial]: Preferences provide a platform-language selector distinct from response language; a complete named language list is not published on this page. Sources: meingpt:meingpt:preferences. - dach_language_quality — Are German, French and Italian UIs complete and natural? [unknown]: Are German, French and Italian UIs complete and natural? Unknown. No authenticated DE/FR/IT/EN UI inspection or WCAG conformance statement was located in the user guide or security pages; admin language completeness is untested. - accessibility_statement — Is there a WCAG conformance statement? [unknown]: Is there a WCAG conformance statement? Unknown. No authenticated DE/FR/IT/EN UI inspection or WCAG conformance statement was located in the user guide or security pages; admin language completeness is untested. - admin_ui_languages — Is the admin interface available in German? [unknown]: Is the admin interface available in German? Unknown. No authenticated DE/FR/IT/EN UI inspection or WCAG conformance statement was located in the user guide or security pages; admin language completeness is untested. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:preferences #### T3 Chat Verification status: No public information found (not-published) Support: not-published We found no public information on this. A missing finding is not evidence that the product lacks the capability. ## Administration and governance ### ADM-01: Single sign-on and authentication Judgment rule: Record SAML, OIDC, Microsoft Entra ID, Google, LDAP, MFA enforcement, conditional-access compatibility, domain verification and plan restrictions. Test one SSO login on a trial where possible. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial Microsoft/Google login is included; custom OIDC is enabled on request and SAML is excluded. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - sso_plan: Platform pricing includes Microsoft/Google/Magic Link; own OIDC needs workspace enablement, while Governance prices Entra group sync and provisioning. Exact custom-OIDC fee is not stated. - sso_protocols — Which SSO protocols and identity providers work? [supported]: Google and Microsoft sign-in plus custom OpenID Connect for Entra, Keycloak, Okta, PingFederate and Authentik. SAML is not supported. Sources: meingpt:meingpt:sso. - sso_plan — Which plan includes SSO? [partial]: Platform pricing includes Microsoft/Google/Magic Link; own OIDC needs workspace enablement, while Governance prices Entra group sync and provisioning. Exact custom-OIDC fee is not stated. Sources: meingpt:meingpt:sso, meingpt:meingpt:prices. - mfa — Can MFA be enforced? [supported]: MFA is enforced at the customer IdP; disabling a user prevents new sign-ins but does not immediately invalidate an existing session. Sources: meingpt:meingpt:sso. - conditional_access — Do Entra conditional-access policies apply? [supported]: Customer IdP controls Conditional Access. Existing sessions expire normally unless separately removing the membership. Sources: meingpt:meingpt:sso. - domain_capture — Can the company claim its email domain? [supported]: DNS TXT verification enables domain auto-join and optional rejection of invitations into other workspaces; existing external memberships remain. Sources: meingpt:meingpt:sso. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:sso, meingpt:meingpt:prices #### T3 Chat Verification status: Agent cross-checked (agent-cross-checked) Support: partial The security policy asserts MFA support without documenting enterprise SAML/OIDC administration. Availability: generally-available Delivery: native Scope: surfaces=T3 Chat browser application; version=Public FAQ, legal pages and vendor replies retrieved 2026-10-21; no authenticated deployment observed; plan=Individual Free, Pro and Premier evidence; no organizational plan or entitlement established; add_ons=Optional customer API keys; plan-specific quotas and checkout details not fully public; deployment_mode=not specified; region=US-hosted service with provider processing in US or other countries; no EU/CH-only option established; os=not specified. Limitations: - Documented findings only; unestablished atomic questions remain explicit gaps. - sso_protocols — Which SSO protocols and identity providers work? [partial]: Security policy refers to OAuth provider accounts and standard authentication without naming enterprise SAML/OIDC protocols or supported corporate IdPs. Sources: t3-chat:t3-chat:security. - sso_plan — Which plan includes SSO? [unknown]: Which plan includes SSO? No business-plan SSO entitlement, Entra conditional-access integration or email-domain capture policy was retrieved. - mfa — Can MFA be enforced? [partial]: The vendor asserts MFA support and recommends two-factor protection at the OAuth provider. Enforceable organization-wide MFA settings were not established. Sources: t3-chat:t3-chat:security. - conditional_access — Do Entra conditional-access policies apply? [unknown]: Do Entra conditional-access policies apply? No business-plan SSO entitlement, Entra conditional-access integration or email-domain capture policy was retrieved. - domain_capture — Can the company claim its email domain? [unknown]: Can the company claim its email domain? No business-plan SSO entitlement, Entra conditional-access integration or email-domain capture policy was retrieved. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: t3-chat:t3-chat:security ### ADM-02: User provisioning and lifecycle Judgment rule: Record SCIM, just-in-time provisioning, group sync from the identity provider, deprovisioning effects (chats, assistants, knowledge ownership transfer) and licence assignment. Deprovision a test user and record what happens to their content. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial Governance includes SCIM; user provisioning and login-time group sync have separate effects. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform plus Governance provisioning add-on; add_ons=Governance; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - group_sync: Entra group-to-team synchronization is separate from user-only SCIM and adds memberships on login. It is not a general bidirectional SCIM-group sync. - deprovisioning: SCIM deactivates membership, preserves data and releases the seat. Shared-asset transfer behavior was not trial-tested. - scim — Is SCIM provisioning supported? [supported]: Entra SCIM v1 creates/updates/deactivates users, requires a verified domain and workspace enablement, and replaces manual invitations. Sources: meingpt:meingpt:scim, meingpt:meingpt:prices. - group_sync — Are identity-provider groups synced? [partial]: Entra group-to-team synchronization is separate from user-only SCIM and adds memberships on login. It is not a general bidirectional SCIM-group sync. Sources: meingpt:meingpt:scim, meingpt:meingpt:sso. - jit_provisioning — Are users created at first login? [supported]: Custom OIDC can join a verified-domain workspace on first authenticated login; plain Microsoft sign-in alone does not grant workspace membership. Sources: meingpt:meingpt:sso, meingpt:meingpt:scim. - deprovisioning — What happens to a removed user's data and shared assets? [partial]: SCIM deactivates membership, preserves data and releases the seat. Shared-asset transfer behavior was not trial-tested. Sources: meingpt:meingpt:scim. - seat_assignment — Can admins assign and reclaim seats? [supported]: New SCIM users start as unlicensed viewers unless automatic seat assignment is on; admins can assign seats. Sources: meingpt:meingpt:scim. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:scim, meingpt:meingpt:prices, meingpt:meingpt:sso #### T3 Chat Verification status: No public information found (not-published) Support: not-published We found no public information on this. A missing finding is not evidence that the product lacks the capability. ### ADM-03: Roles, groups and feature permissions Judgment rule: Record admin roles, delegated admins, groups, per-feature permissions (web search, image generation, connectors, file upload, agents), separation of subsidiaries or departments, and enforcement. Test one disabled feature for one group. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial Role and group controls include request-enabled app permissions and model restrictions. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - multi_entity: Multiple workspaces can represent separate companies, but each has separate credit and billing; a unified multi-entity contract was not established. - admin_roles — Which admin roles exist, and can admin rights be delegated? [supported]: Admin, Member and Viewer; unlimited workspace admins and separate team administrators are documented. Workspace admins consume a seat and bypass restrictions. Sources: meingpt:meingpt:roles, meingpt:meingpt:permissions. - groups — Can permissions be set per group? [supported]: Teams control shared resources and models; admins see all team-shared assets. Sources: meingpt:meingpt:roles, meingpt:meingpt:config. - feature_toggles — Which features can be switched on or off per group? [supported]: Request-enabled controls restrict image, meeting notes/bot, translation, web, Voice Agent, Buddy and automation use by teams or persons. Sources: meingpt:meingpt:permissions. - multi_entity — Can subsidiaries or departments be separated within one contract? [partial]: Multiple workspaces can represent separate companies, but each has separate credit and billing; a unified multi-entity contract was not established. Sources: meingpt:meingpt:sso. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:roles, meingpt:meingpt:permissions, meingpt:meingpt:config, meingpt:meingpt:sso #### T3 Chat Verification status: No public information found (not-published) Support: not-published We found no public information on this. A missing finding is not evidence that the product lacks the capability. ### ADM-04: Usage and adoption analytics Judgment rule: Record dashboards for active users, usage by feature, model, group and assistant, export, and privacy controls (aggregation, pseudonymisation, works-council-friendly settings). Distinguish aggregate adoption metrics from per-person monitoring. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial Adoption metrics and cost CSVs have team and anonymization options. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - breakdowns: Usage-based plans report costs by user, team, model and platform/API; complete assistant/feature breakdown is not demonstrated. - adoption_dashboard — Is there an adoption dashboard? [supported]: Cockpit lists prompts, active users, daily prompts/workflow runs and popular workflows. Sources: meingpt:meingpt:analytics. - breakdowns — Can usage be broken down by group, model, feature and assistant? [partial]: Usage-based plans report costs by user, team, model and platform/API; complete assistant/feature breakdown is not demonstrated. Sources: meingpt:meingpt:analytics. - export — Can usage data be exported? [supported]: User/team cost statistics export to CSV. Sources: meingpt:meingpt:analytics. - privacy_modes — Can per-person reporting be limited or pseudonymised? [supported]: Metric anonymization replaces names with IDs; top-user metrics depend on that setting. Sources: meingpt:meingpt:analytics. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:analytics #### T3 Chat Verification status: No public information found (not-published) Support: not-published We found no public information on this. A missing finding is not evidence that the product lacks the capability. ### ADM-05: Spend controls and cost allocation Judgment rule: Record cost visibility per user, group and cost centre, budgets and their behaviour (alert, block, downgrade model), pooled credits, and chargeback exports. Test a small budget on a trial where possible. Distinguish billed amounts, estimates and unavailable amounts. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial Personal category budgets and pooled credit provide controls with documented exceptions. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - budget_enforcement: Zero blocks new category use and API exhaustion returns 403; running meeting bots can overshoot and automated runs are outside personal categories. - chargeback: CSV cost exports exist by user/team; formal cost-centre coding is not established. - cost_visibility — Can admins see cost or consumption per user and group? [supported]: Usage-based plans show EUR costs per user/model and team, separately for platform and API. Sources: meingpt:meingpt:analytics. - budgets — Can budgets be set per user or group? [supported]: Six personal usage levels configure monthly budgets across five categories and are assigned singly or in bulk. Sources: meingpt:meingpt:limits. - budget_enforcement — Does reaching a budget alert, block or downgrade? [partial]: Zero blocks new category use and API exhaustion returns 403; running meeting bots can overshoot and automated runs are outside personal categories. Sources: meingpt:meingpt:limits, meingpt:meingpt:api. - chargeback — Can costs be exported per cost centre? [partial]: CSV cost exports exist by user/team; formal cost-centre coding is not established. Sources: meingpt:meingpt:analytics. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:analytics, meingpt:meingpt:limits, meingpt:meingpt:api #### T3 Chat Verification status: Agent cross-checked (agent-cross-checked) Support: partial Individual consumption indicators exist; organization budgets and chargeback remain unverified. Availability: generally-available Delivery: native Scope: surfaces=T3 Chat browser application; version=Public FAQ, legal pages and vendor replies retrieved 2026-10-21; no authenticated deployment observed; plan=Individual Pro USD8 and Premier account consumption mechanics; no organization/admin budget offer asserted; add_ons=Optional customer API keys; plan-specific quotas and checkout details not fully public; deployment_mode=not specified; region=US-hosted service with provider processing in US or other countries; no EU/CH-only option established; os=not specified. Limitations: - Documented findings only; unestablished atomic questions remain explicit gaps. - No complete Free/Pro/Premier entitlement matrix or authenticated model/feature menu was retrieved. - cost_visibility — Can admins see cost or consumption per user and group? [partial]: The paid Pro discussion and Premier support reply describe per-account cost/bucket consumption. The FAQ describes separate balance displays; exact placement and model-cost indicators are not a customer organization reporting entitlement. Sources: t3-chat:t3-chat:faq-usage, t3-chat:t3-chat:r2-paid-tier-controls, t3-chat:t3-chat:r2-premier-buckets. - budgets — Can budgets be set per user or group? [unknown]: Can budgets be set per user or group? No customer organization/user-group budget assignment or cost-centre export was retrieved. - budget_enforcement — Does reaching a budget alert, block or downgrade? [partial]: For the documented paid-account bucket routes, exhausting both balances requires waiting for refill. This does not establish configurable user/group currency budgets, alerts or downgrade policies. Sources: t3-chat:t3-chat:faq-usage, t3-chat:t3-chat:r2-premier-buckets. - chargeback — Can costs be exported per cost centre? [unknown]: Can costs be exported per cost centre? No customer organization/user-group budget assignment or cost-centre export was retrieved. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: t3-chat:t3-chat:faq-usage, t3-chat:t3-chat:r2-paid-tier-controls, t3-chat:t3-chat:r2-premier-buckets ### ADM-06: Audit logs and compliance export Judgment rule: Record which events are logged (logins, admin changes, prompts, responses, file access, connector queries), retention, export or SIEM streaming, and eDiscovery/legal hold. Use the word audit only with a defined retention and integrity model. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial Administrative change exports and a separate Outpost search log have distinct retention. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - admin_events: Administrative settings and SCIM changes are logged with actor and changes; a login/security-event coverage inventory and tamper-evidence model were not found. - content_logging: The settings-change log explicitly excludes conversation contents; a separate API-logging setting exists but prompt/response compliance-retention coverage is unclear. - siem_export: CSV/JSON filtered audit export, 10000 entries/20 MiB/two minutes per export; SIEM streaming is not established. - Unverified ediscovery: No legal-hold or eDiscovery documentation was located; immutable audit storage and full content-logging configuration were not established. - admin_events — Are admin and security events logged? [partial]: Administrative settings and SCIM changes are logged with actor and changes; a login/security-event coverage inventory and tamper-evidence model were not found. Sources: meingpt:meingpt:audit, meingpt:meingpt:scim. - content_logging — Can prompts and answers be retained for compliance, and is it configurable? [partial]: The settings-change log explicitly excludes conversation contents; a separate API-logging setting exists but prompt/response compliance-retention coverage is unclear. Sources: meingpt:meingpt:audit. - siem_export — Can logs stream to a SIEM or be exported? [partial]: CSV/JSON filtered audit export, 10000 entries/20 MiB/two minutes per export; SIEM streaming is not established. Sources: meingpt:meingpt:audit-export. - ediscovery — Are eDiscovery and legal hold supported? [unknown]: Are eDiscovery and legal hold supported? Unknown. No legal-hold or eDiscovery documentation was located; immutable audit storage and full content-logging configuration were not established. - log_retention — How long are logs kept? [supported]: Administrative settings changes are retained indefinitely; Governance Outpost query records are retained 90 days from enablement. Sources: meingpt:meingpt:audit. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:audit, meingpt:meingpt:scim, meingpt:meingpt:audit-export #### T3 Chat Verification status: Agent cross-checked (agent-cross-checked) Support: partial The privacy policy describes vendor telemetry, not a customer compliance logging service. Availability: generally-available Delivery: native Scope: surfaces=T3 Chat browser application; version=Public FAQ, legal pages and vendor replies retrieved 2026-10-21; no authenticated deployment observed; plan=Individual Free, Pro and Premier evidence; no organizational plan or entitlement established; add_ons=Optional customer API keys; plan-specific quotas and checkout details not fully public; deployment_mode=not specified; region=US-hosted service with provider processing in US or other countries; no EU/CH-only option established; os=not specified. Limitations: - Documented findings only; unestablished atomic questions remain explicit gaps. - admin_events — Are admin and security events logged? [unknown]: Are admin and security events logged? No customer admin/security-event catalogue, SIEM stream/export, eDiscovery/legal hold or fixed log retention was retrieved. Vendor monitoring in the security policy is not evidence of customer audit access. - content_logging — Can prompts and answers be retained for compliance, and is it configurable? [partial]: The privacy policy collects prompts/uploads and Terms reserve review/removal for violations. A configurable customer prompt/answer compliance archive was not established. Sources: t3-chat:t3-chat:privacy, t3-chat:t3-chat:terms. - siem_export — Can logs stream to a SIEM or be exported? [unknown]: Can logs stream to a SIEM or be exported? No customer admin/security-event catalogue, SIEM stream/export, eDiscovery/legal hold or fixed log retention was retrieved. Vendor monitoring in the security policy is not evidence of customer audit access. - ediscovery — Are eDiscovery and legal hold supported? [unknown]: Are eDiscovery and legal hold supported? No customer admin/security-event catalogue, SIEM stream/export, eDiscovery/legal hold or fixed log retention was retrieved. Vendor monitoring in the security policy is not evidence of customer audit access. - log_retention — How long are logs kept? [unknown]: How long are logs kept? No customer admin/security-event catalogue, SIEM stream/export, eDiscovery/legal hold or fixed log retention was retrieved. Vendor monitoring in the security policy is not evidence of customer audit access. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: t3-chat:t3-chat:privacy, t3-chat:t3-chat:terms ### ADM-07: Data loss prevention and personal data Judgment rule: Record detection and handling of personal and confidential data before it reaches a model: redaction, masking, blocking, warning. Record integration with Purview, Netskope or similar, and whether filtering runs locally or at a third party. Send a fixture prompt with a Swiss AHV number and IBAN and record the outcome. #### meinGPT Verification status: No public information found (not-published) Support: not-published We found no public information on this. A missing finding is not evidence that the product lacks the capability. #### T3 Chat Verification status: No public information found (not-published) Support: not-published We found no public information on this. A missing finding is not evidence that the product lacks the capability. ### ADM-08: AI policy and guardrails Judgment rule: Record organisation-wide system instructions, acceptable-use notices and acknowledgement, content filters, topic restrictions, AI-output labelling, and admin visibility of policy violations. Separate enforced policy from instructions the model is expected to follow. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial Workspace instructions and Governance acknowledgements are documented; policy enforcement breadth is limited. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - org_instructions: Additional Chat Context is a workspace-wide system instruction, not a guaranteed enforcement layer. - content_filters: Provider content filtering can stop output; configurable organization-wide topic filters were not established. - Unverified violation_reporting: Admin policy-violation reporting was not established in Compliance Controls; no guardrail fixture was run. - org_instructions — Can admins set organisation-wide instructions? [partial]: Additional Chat Context is a workspace-wide system instruction, not a guaranteed enforcement layer. Sources: meingpt:meingpt:config. - usage_notice — Can users be required to accept a usage policy? [supported]: Governance includes multilingual acceptable-use policies for acknowledgement and mandatory courses that block other pages until complete. Sources: meingpt:meingpt:prices, meingpt:meingpt:audit. - content_filters — Are content filters configurable? [partial]: Provider content filtering can stop output; configurable organization-wide topic filters were not established. Sources: meingpt:meingpt:api. - violation_reporting — Are policy hits visible to admins? [unknown]: Are policy hits visible to admins? Unknown. Admin policy-violation reporting was not established in Compliance Controls; no guardrail fixture was run. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:config, meingpt:meingpt:prices, meingpt:meingpt:audit, meingpt:meingpt:api #### T3 Chat Verification status: Agent cross-checked (agent-cross-checked) Support: partial Acceptable-use terms apply, while organization policy controls remain unverified. Availability: generally-available Delivery: native Scope: surfaces=T3 Chat browser application; version=Public FAQ, legal pages and vendor replies retrieved 2026-10-21; no authenticated deployment observed; plan=Individual Free, Pro and Premier evidence; no organizational plan or entitlement established; add_ons=Optional customer API keys; plan-specific quotas and checkout details not fully public; deployment_mode=not specified; region=US-hosted service with provider processing in US or other countries; no EU/CH-only option established; os=not specified. Limitations: - Documented findings only; unestablished atomic questions remain explicit gaps. - org_instructions — Can admins set organisation-wide instructions? [unknown]: Can admins set organisation-wide instructions? No admin global-instruction policy, mandatory company policy acknowledgement, configurable deterministic filter or customer violation-hit reporting was retrieved. - usage_notice — Can users be required to accept a usage policy? [unknown]: Can users be required to accept a usage policy? No admin global-instruction policy, mandatory company policy acknowledgement, configurable deterministic filter or customer violation-hit reporting was retrieved. - content_filters — Are content filters configurable? [partial]: Terms prohibit categories of unlawful/harmful content and permit vendor review/removal or suspension. Customer configurable content filters are not established. Sources: t3-chat:t3-chat:terms. - violation_reporting — Are policy hits visible to admins? [unknown]: Are policy hits visible to admins? No admin global-instruction policy, mandatory company policy acknowledgement, configurable deterministic filter or customer violation-hit reporting was retrieved. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: t3-chat:t3-chat:terms ## Data protection and sovereignty ### DATA-01: Hosting location and data residency Judgment rule: Record country or region separately for chat storage, files and indexes, model inference, backups, logs and telemetry, and support access. Name the cloud provider. Record whether residency is default, a paid option or plan-dependent. "EU data boundary" programmes are recorded with their documented exceptions. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial Germany platform storage and model-specific EU/global routes must be scoped separately. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - inference_region: Azure Sweden/France, Vertex Frankfurt, Mistral France/Sweden, OpenAI EU endpoint, Nebius Finland, TensorX Ireland and other conditional routes. Some image variants remain Global; the accepted DPA governs authorized exceptions. - backups_logs_region: Scaleway France is conditional backup infrastructure; log/telemetry locations are not separately specified. - residency_plan: DPA 2.0.1 governs new acceptance since 24 September; existing customers retain agreed versions. EU inference depends on selected/approved routes; section 4(3) permits listed/instructed third-country transfers. - Unverified support_access_region, swiss_hosting: No Swiss hosting entitlement or complete geographical support-access and log/telemetry matrix was found. - storage_region — Where are chats, files and indexes stored? [supported]: Hetzner Germany hosts chats, accounts, documents and cloud indexes; register identifies Nürnberg/Falkenstein. Sources: meingpt:meingpt:pools, meingpt:meingpt:subprocessors. - inference_region — Where do model calls run, per provider? [partial]: Azure Sweden/France, Vertex Frankfurt, Mistral France/Sweden, OpenAI EU endpoint, Nebius Finland, TensorX Ireland and other conditional routes. Some image variants remain Global; the accepted DPA governs authorized exceptions. Sources: meingpt:meingpt:subprocessors, meingpt:meingpt:image-admin, meingpt:meingpt:dpa. - backups_logs_region — Where are backups, logs and telemetry? [partial]: Scaleway France is conditional backup infrastructure; log/telemetry locations are not separately specified. Sources: meingpt:meingpt:subprocessors, meingpt:meingpt:dpa. - support_access_region — From where can vendor staff access data? [unknown]: From where can vendor staff access data? Unknown. No Swiss hosting entitlement or complete geographical support-access and log/telemetry matrix was found. - swiss_hosting — Is Swiss hosting available? [unknown]: Is Swiss hosting available? Unknown. No Swiss hosting entitlement or complete geographical support-access and log/telemetry matrix was found. - residency_plan — Which plan or option provides residency? [partial]: DPA 2.0.1 governs new acceptance since 24 September; existing customers retain agreed versions. EU inference depends on selected/approved routes; section 4(3) permits listed/instructed third-country transfers. Sources: meingpt:meingpt:dpa-page, meingpt:meingpt:dpa. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:pools, meingpt:meingpt:subprocessors, meingpt:meingpt:image-admin, meingpt:meingpt:dpa, meingpt:meingpt:dpa-page #### T3 Chat Verification status: Agent cross-checked (agent-cross-checked) Support: partial The privacy policy places the service under US/international processing; no EU/Swiss end-to-end option was established. Availability: generally-available Delivery: native Scope: surfaces=T3 Chat browser application; version=Public FAQ, legal pages and vendor replies retrieved 2026-10-21; no authenticated deployment observed; plan=Individual Free, Pro and Premier evidence; no organizational plan or entitlement established; add_ons=Optional customer API keys; plan-specific quotas and checkout details not fully public; deployment_mode=not specified; region=US-hosted service with provider processing in US or other countries; no EU/CH-only option established; os=not specified. Limitations: - Documented findings only; unestablished atomic questions remain explicit gaps. - storage_region — Where are chats, files and indexes stored? [partial]: The policy says services are provided in the United States and information is processed there or in other countries; Azure, AWS and Vercel are named. Component-specific chat/file/index storage regions are not enumerated. Sources: t3-chat:t3-chat:privacy. - inference_region — Where do model calls run, per provider? [partial]: The policy names OpenAI, Google, Anthropic and OpenRouter processing, but no provider-by-provider region, EU-only switch or customer region guarantee is supplied. Sources: t3-chat:t3-chat:privacy, t3-chat:t3-chat:faq-training. - backups_logs_region — Where are backups, logs and telemetry? [unknown]: Where are backups, logs and telemetry? No exact backup/log/telemetry country matrix, support-access location, Swiss hosting offer or residency-plan entitlement was retrieved. - support_access_region — From where can vendor staff access data? [unknown]: From where can vendor staff access data? No exact backup/log/telemetry country matrix, support-access location, Swiss hosting offer or residency-plan entitlement was retrieved. - swiss_hosting — Is Swiss hosting available? [unknown]: Is Swiss hosting available? No exact backup/log/telemetry country matrix, support-access location, Swiss hosting offer or residency-plan entitlement was retrieved. - residency_plan — Which plan or option provides residency? [unknown]: Which plan or option provides residency? No exact backup/log/telemetry country matrix, support-access location, Swiss hosting offer or residency-plan entitlement was retrieved. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: t3-chat:t3-chat:privacy, t3-chat:t3-chat:faq-training ### DATA-02: Deployment options Judgment rule: Record multi-tenant SaaS, dedicated single-tenant instance, deployment in the customer's cloud account, on-premise and air-gapped as separate options with plan and availability. Record which features are missing in each mode, who operates it, and the documented seat or contract threshold for each mode, which can differ from the enterprise plan threshold (e.g. dedicated SaaS from 2,000 seats, own cloud from 5,000). A local-network connector or gateway that reaches on-prem data from a vendor-hosted platform is KNOW-03, not on-premise deployment. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial SaaS is offered; terms permit individually agreed on-premise use, while Outpost is a connector. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace, meinGPT Outpost local connector; version=Public documentation checked 2026-10-21; plan=Public SaaS; on-premise only if specifically contracted; add_ons=none stated; deployment_mode=not specified; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - on_premise: Terms section 1a permits the whole software in customer data centres when agreed in the licence contract. The linked on-premise-requirements URL redirects to the generic security page; minimum seats and setup/operator terms are not public. - mode_feature_gaps: The code sandbox remains in Frankfurt cloud even for on-premise customers; there is no complete mode parity list. - Unverified dedicated_instance, customer_cloud, air_gapped: Dedicated-instance, customer-cloud and air-gapped entitlements, operators, minimum seats and prices were not established. Outpost is a local connector, not evidence for those modes. - multi_tenant_saas — Is a shared SaaS offered? [supported]: The public web workspace and shared organization API limits establish the managed SaaS offer. Sources: meingpt:meingpt:terms, meingpt:meingpt:api. - dedicated_instance — Is a dedicated single-tenant instance offered, from what size? [unknown]: Is a dedicated single-tenant instance offered, from what size? Unknown. Dedicated-instance, customer-cloud and air-gapped entitlements, operators, minimum seats and prices were not established. Outpost is a local connector, not evidence for those modes. - customer_cloud — Can it run in the customer's own cloud account, from what size? [unknown]: Can it run in the customer's own cloud account, from what size? Unknown. Dedicated-instance, customer-cloud and air-gapped entitlements, operators, minimum seats and prices were not established. Outpost is a local connector, not evidence for those modes. - on_premise — Can the whole platform run on customer infrastructure, from what size? [partial]: Terms section 1a permits the whole software in customer data centres when agreed in the licence contract. The linked on-premise-requirements URL redirects to the generic security page; minimum seats and setup/operator terms are not public. Sources: meingpt:meingpt:terms. - air_gapped — Can it run without internet access? [unknown]: Can it run without internet access? Unknown. Dedicated-instance, customer-cloud and air-gapped entitlements, operators, minimum seats and prices were not established. Outpost is a local connector, not evidence for those modes. - mode_feature_gaps — Which features are missing outside SaaS? [partial]: The code sandbox remains in Frankfurt cloud even for on-premise customers; there is no complete mode parity list. Sources: meingpt:meingpt:sandbox. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:terms, meingpt:meingpt:api, meingpt:meingpt:sandbox #### T3 Chat Verification status: Agent cross-checked (agent-cross-checked) Support: partial A hosted web service is offered; tenant architecture and alternative deployment contracts remain unverified. Availability: generally-available Delivery: native Scope: surfaces=T3 Chat browser application; version=Public FAQ, legal pages and vendor replies retrieved 2026-10-21; no authenticated deployment observed; plan=Individual Free, Pro and Premier evidence; no organizational plan or entitlement established; add_ons=Optional customer API keys; plan-specific quotas and checkout details not fully public; deployment_mode=not specified; region=US-hosted service with provider processing in US or other countries; no EU/CH-only option established; os=not specified. Limitations: - Documented findings only; unestablished atomic questions remain explicit gaps. - multi_tenant_saas — Is a shared SaaS offered? [partial]: The public web service accepts individual accounts and processes content with hosting/model providers. Its exact multitenant architecture and boundaries are not documented in the retrieved public material. Sources: t3-chat:t3-chat:terms, t3-chat:t3-chat:privacy. - dedicated_instance — Is a dedicated single-tenant instance offered, from what size? [unknown]: Is a dedicated single-tenant instance offered, from what size? No vendor-operated dedicated tenant, customer-cloud/on-premise package, offline licence or deployment-specific feature matrix was retrieved. Independent T3 clones and the distinct T3 Code product do not establish a self-host licence. - customer_cloud — Can it run in the customer's own cloud account, from what size? [unknown]: Can it run in the customer's own cloud account, from what size? No vendor-operated dedicated tenant, customer-cloud/on-premise package, offline licence or deployment-specific feature matrix was retrieved. Independent T3 clones and the distinct T3 Code product do not establish a self-host licence. - on_premise — Can the whole platform run on customer infrastructure, from what size? [unknown]: Can the whole platform run on customer infrastructure, from what size? No vendor-operated dedicated tenant, customer-cloud/on-premise package, offline licence or deployment-specific feature matrix was retrieved. Independent T3 clones and the distinct T3 Code product do not establish a self-host licence. - air_gapped — Can it run without internet access? [unknown]: Can it run without internet access? No vendor-operated dedicated tenant, customer-cloud/on-premise package, offline licence or deployment-specific feature matrix was retrieved. Independent T3 clones and the distinct T3 Code product do not establish a self-host licence. - mode_feature_gaps — Which features are missing outside SaaS? [unknown]: Which features are missing outside SaaS? No vendor-operated dedicated tenant, customer-cloud/on-premise package, offline licence or deployment-specific feature matrix was retrieved. Independent T3 clones and the distinct T3 Code product do not establish a self-host licence. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: t3-chat:t3-chat:terms, t3-chat:t3-chat:privacy ### DATA-03: Training use and provider retention Judgment rule: Record whether customer content trains vendor or provider models, by plan and default; zero-data-retention arrangements with model providers; abuse-monitoring retention and human review; and the model provider's terms that apply. A "no training" claim on a marketing page is `vendor-claimed` until the contract or DPA says so. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial DPA 2.0.1 prohibits training and marks conditional AI services for no permanent provider content copy. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - provider_retention: Annex 4 defines ZDR as no permanent copy on the provider infrastructure; exact abuse-monitoring time and exceptions are not specified. - zero_retention: Conditional AI/function services in Annex 4 carry ZDR, while SelectCode-controlled storage remains separate; applies to the accepted DPA version. - Unverified human_review: Human provider-review rights and per-model transient/abuse-retention durations are not disclosed in the accepted-form DPA or data-flow guide. - vendor_training — Does the vendor train on customer content? [supported]: Section 5(5) restricts processing to service delivery and excludes other purposes. Sources: meingpt:meingpt:dpa. - provider_training — Do model providers train on it under this product's contracts? [supported]: Section 5(5) states no subprocessor uses customer data to train its own AI model. Sources: meingpt:meingpt:dpa. - provider_retention — How long do model providers keep prompts, including for abuse monitoring? [partial]: Annex 4 defines ZDR as no permanent copy on the provider infrastructure; exact abuse-monitoring time and exceptions are not specified. Sources: meingpt:meingpt:dpa. - zero_retention — Is zero data retention available, and on which plan? [partial]: Conditional AI/function services in Annex 4 carry ZDR, while SelectCode-controlled storage remains separate; applies to the accepted DPA version. Sources: meingpt:meingpt:dpa. - human_review — Can provider staff review content? [unknown]: Can provider staff review content? Unknown. Human provider-review rights and per-model transient/abuse-retention durations are not disclosed in the accepted-form DPA or data-flow guide. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:dpa #### T3 Chat Verification status: Agent cross-checked (agent-cross-checked) Support: partial No vendor training is asserted, with provider exceptions and customer-key responsibility explicitly retained. Availability: generally-available Delivery: native Scope: surfaces=T3 Chat browser application; version=Public FAQ, legal pages and vendor replies retrieved 2026-10-21; no authenticated deployment observed; plan=Individual Free, Pro and Premier evidence; no organizational plan or entitlement established; add_ons=Optional customer API keys; plan-specific quotas and checkout details not fully public; deployment_mode=not specified; region=US-hosted service with provider processing in US or other countries; no EU/CH-only option established; os=not specified. Limitations: - Documented findings only; unestablished atomic questions remain explicit gaps. - vendor_training — Does the vendor train on customer content? [supported]: FAQ says T3 does not train its own models and the privacy policy says personal information is not used to train AI models. Sources: t3-chat:t3-chat:faq-training, t3-chat:t3-chat:privacy. - provider_training — Do model providers train on it under this product's contracts? [supported]: The FAQ says OpenAI/Anthropic defaults are not opted into training, Google paid models disable training and OpenRouter training routes are disabled. Some new models may require training and are labeled; BYOK customers manage their own provider settings. Sources: t3-chat:t3-chat:faq-training. - provider_retention — How long do model providers keep prompts, including for abuse monitoring? [unknown]: How long do model providers keep prompts, including for abuse monitoring? The FAQ does not establish provider prompt/abuse-monitoring retention, uniform ZDR entitlement or all provider human-review exceptions. No universal no-training/no-retention promise is inferred. - zero_retention — Is zero data retention available, and on which plan? [unknown]: Is zero data retention available, and on which plan? The FAQ does not establish provider prompt/abuse-monitoring retention, uniform ZDR entitlement or all provider human-review exceptions. No universal no-training/no-retention promise is inferred. - human_review — Can provider staff review content? [unknown]: Can provider staff review content? The FAQ does not establish provider prompt/abuse-monitoring retention, uniform ZDR entitlement or all provider human-review exceptions. No universal no-training/no-retention promise is inferred. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: t3-chat:t3-chat:faq-training, t3-chat:t3-chat:privacy ### DATA-04: Retention and deletion Judgment rule: Record default retention for chats, files, indexes and logs; admin-configured retention; user deletion; deletion after contract end; and deletion confirmation. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial Ordinary chat removal archives; admin retention and contractual deletion use different mechanisms. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - user_deletion: Chat removal does not permanently delete messages or attachments; final account deletion is requested by the administrator via support. Generated images/text translations have their own deletion controls. - default_retention — What is retained and for how long by default? [supported]: Chat/file data stays until instructed deletion; UI Chat löschen archives. Text translations remain until user deletion; document translations persist until account/workspace deletion. Sources: meingpt:meingpt:flow, meingpt:meingpt:translation. - admin_retention — Can admins set retention periods? [supported]: Release 2.67.0 adds disabled-by-default deletion periods for chats, meetings and automation runs, with a 24-hour wait before a new/shorter period applies. Sources: meingpt:meingpt:changes. - user_deletion — Can users delete their chats and files permanently? [partial]: Chat removal does not permanently delete messages or attachments; final account deletion is requested by the administrator via support. Generated images/text translations have their own deletion controls. Sources: meingpt:meingpt:flow, meingpt:meingpt:security, meingpt:meingpt:images, meingpt:meingpt:translation. - contract_end_deletion — What is deleted at contract end, and when? [supported]: DPA section 12 requires return or requested deletion, including backups, within 30 days after termination or an agreed export window; statutory retention exceptions apply and proof is available on request. Sources: meingpt:meingpt:dpa. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:flow, meingpt:meingpt:translation, meingpt:meingpt:changes, meingpt:meingpt:security, meingpt:meingpt:images, meingpt:meingpt:dpa #### T3 Chat Verification status: Agent cross-checked (agent-cross-checked) Support: partial Termination export responsibilities are contractual; plan-specific temporary-chat behaviour and retention remain unknown. Availability: generally-available Delivery: native Scope: surfaces=T3 Chat browser application; version=Public FAQ, legal pages and vendor replies retrieved 2026-10-21; no authenticated deployment observed; plan=Individual T3 Chat service Terms; temporary-chat Free/Pro/Premier entitlement not established; add_ons=Optional customer API keys; plan-specific quotas and checkout details not fully public; deployment_mode=not specified; region=US-hosted service with provider processing in US or other countries; no EU/CH-only option established; os=not specified. Limitations: - Documented findings only; unestablished atomic questions remain explicit gaps. - No complete Free/Pro/Premier entitlement matrix or authenticated model/feature menu was retrieved. - default_retention — What is retained and for how long by default? [unknown]: The generic URL FAQ distinguishes temporary and persisted threads but does not map that feature to a named individual tier or specify server/provider/file/backup retention. No plan-specific temporary-chat retention claim is made. - admin_retention — Can admins set retention periods? [unknown]: Can admins set retention periods? No administrator retention controls, user permanent-deletion completion guarantee, backup erasure interval or fixed contract-end deletion deadline was retrieved. - user_deletion — Can users delete their chats and files permanently? [unknown]: Can users delete their chats and files permanently? No administrator retention controls, user permanent-deletion completion guarantee, backup erasure interval or fixed contract-end deletion deadline was retrieved. - contract_end_deletion — What is deleted at contract end, and when? [partial]: Terms put responsibility on users to export before termination and endeavor to provide a reasonable retrieval opportunity only for vendor-convenience termination; deletion timing and certificate are unspecified. Sources: t3-chat:t3-chat:terms. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: t3-chat:t3-chat:terms ### DATA-05: Sub-processors and foreign-law exposure Judgment rule: Record the published sub-processor list with countries, change-notification rules, the vendor's and parent company's jurisdiction, transfer mechanisms (EU-US Data Privacy Framework, Swiss-US DPF, SCCs) and the vendor's documented position on foreign government access requests (e.g. US CLOUD Act). Record facts, not a legal assessment. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial DPA and processing register name providers, conditional use and advance change notices. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - parent_jurisdiction: SelectCode GmbH is a German entity; the security site describes it as independent. A separately owned parent was not identified. - subprocessor_list — Is a dated sub-processor list published? [supported]: DPA Annex 4 is dated 24 September 2026; live function/region register is also public. Sources: meingpt:meingpt:dpa, meingpt:meingpt:subprocessors. - subprocessor_countries — Which countries do sub-processors operate in? [supported]: Contract entities include Germany, Netherlands, Slovenia, Luxembourg, France, Ireland and Finland. Processing register names Germany platform, France backups, Finland, Ireland, Sweden/France and Frankfurt inference routes; EU symbol is processing, not ownership. Sources: meingpt:meingpt:dpa, meingpt:meingpt:subprocessors. - change_notification — Are customers notified of sub-processor changes? [supported]: Section 9 requires notice before additions/replacements and gives two weeks to object on important data-protection grounds, with specified termination rights. Sources: meingpt:meingpt:dpa. - parent_jurisdiction — Which jurisdiction governs the vendor and its parent? [partial]: SelectCode GmbH is a German entity; the security site describes it as independent. A separately owned parent was not identified. Sources: meingpt:meingpt:terms, meingpt:meingpt:subprocessors. - transfer_mechanism — Which transfer mechanism applies to non-EU/CH processing? [supported]: DPA section 4(3) uses GDPR Article 45 adequacy or Article 46 safeguards for disclosed/instructed transfers; it does not promise a blanket absence of third-country processing. Sources: meingpt:meingpt:dpa. - government_access_policy — Is there a published government-access or transparency policy? [supported]: Section 7 requires notification where lawful, review/challenge of disproportionate requests, minimization and no compliance with unrecognized third-country requests. Sources: meingpt:meingpt:dpa. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:dpa, meingpt:meingpt:subprocessors, meingpt:meingpt:terms #### T3 Chat Verification status: Agent cross-checked (agent-cross-checked) Support: partial Named providers and a US vendor address are public, without a complete processor-country or transfer contract matrix. Availability: generally-available Delivery: native Scope: surfaces=T3 Chat browser application; version=Public FAQ, legal pages and vendor replies retrieved 2026-10-21; no authenticated deployment observed; plan=Individual Free, Pro and Premier evidence; no organizational plan or entitlement established; add_ons=Optional customer API keys; plan-specific quotas and checkout details not fully public; deployment_mode=not specified; region=US-hosted service with provider processing in US or other countries; no EU/CH-only option established; os=not specified. Limitations: - Documented findings only; unestablished atomic questions remain explicit gaps. - subprocessor_list — Is a dated sub-processor list published? [partial]: The May31,2025 privacy policy names Azure, AWS, Vercel, OpenAI, Google, Anthropic and OpenRouter, explicitly non-exhaustively. This is not established as a complete Article28 subprocessor schedule. Sources: t3-chat:t3-chat:privacy. - subprocessor_countries — Which countries do sub-processors operate in? [unknown]: Which countries do sub-processors operate in? No exhaustive processor operation-country matrix, signed SCC/Swiss transfer mechanism or subprocessor objection schedule was retrieved. - change_notification — Are customers notified of sub-processor changes? [partial]: Privacy policy says material changes are posted with an updated effective date. A processor-addition notice period and objection right are not specified. Sources: t3-chat:t3-chat:privacy. - parent_jurisdiction — Which jurisdiction governs the vendor and its parent? [partial]: The policy and Terms identify T3 Tools Inc with a San Francisco address and US/California legal context. A separate parent entity, incorporation jurisdiction and ownership were not independently established. Sources: t3-chat:t3-chat:privacy, t3-chat:t3-chat:terms. - transfer_mechanism — Which transfer mechanism applies to non-EU/CH processing? [unknown]: Which transfer mechanism applies to non-EU/CH processing? No exhaustive processor operation-country matrix, signed SCC/Swiss transfer mechanism or subprocessor objection schedule was retrieved. - government_access_policy — Is there a published government-access or transparency policy? [partial]: The policy allows disclosures for subpoenas, warrants and legal investigations. No separate transparency report or detailed challenge/notification policy was retrieved. Sources: t3-chat:t3-chat:privacy. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: t3-chat:t3-chat:privacy, t3-chat:t3-chat:terms ### DATA-06: Contracts and regulatory coverage Judgment rule: Record DPA availability and form (click-through, negotiable), reference to Swiss revDSG and GDPR, governing law and court, EU AI Act documentation (provider or deployer role, transparency materials), and sector-specific materials (FINMA outsourcing, banking secrecy, public sector). No compliance verdict. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial A public GDPR DPA and German-law terms are available; sector coverage needs qualification. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - dpa: Version 2.0.1 is a public signed-form or electronic-acceptance DPA; negotiability is not promised. - revdsg_gdpr: GDPR Article 28 governs the DPA; an explicit Swiss revDSG annex was not found. - ai_act_materials: Governance includes usage policies and mandatory training; automated EU AI Act use-case checks are labelled Coming soon, so a complete current provider/deployer dossier was not established. - sector_materials: Governance offers a separate section 203 German professional-secrecy agreement; FINMA-specific materials were not found. - dpa — Is a DPA available, and is it negotiable? [partial]: Version 2.0.1 is a public signed-form or electronic-acceptance DPA; negotiability is not promised. Sources: meingpt:meingpt:dpa-page, meingpt:meingpt:dpa. - revdsg_gdpr — Does the DPA cover revDSG and GDPR? [partial]: GDPR Article 28 governs the DPA; an explicit Swiss revDSG annex was not found. Sources: meingpt:meingpt:dpa. - governing_law — Which law and court govern the contract? [supported]: DPA section 13 uses German law; terms choose the competent court at the vendor seat. Sources: meingpt:meingpt:dpa, meingpt:meingpt:terms. - ai_act_materials — Is EU AI Act documentation provided? [partial]: Governance includes usage policies and mandatory training; automated EU AI Act use-case checks are labelled Coming soon, so a complete current provider/deployer dossier was not established. Sources: meingpt:meingpt:prices. - sector_materials — Are FINMA or other sector materials provided? [partial]: Governance offers a separate section 203 German professional-secrecy agreement; FINMA-specific materials were not found. Sources: meingpt:meingpt:prices. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:dpa-page, meingpt:meingpt:dpa, meingpt:meingpt:terms, meingpt:meingpt:prices #### T3 Chat Verification status: Agent cross-checked (agent-cross-checked) Support: partial Public consumer/service terms use California law; an enterprise GDPR/Swiss DPA remains unverified. Availability: generally-available Delivery: native Scope: surfaces=T3 Chat browser application; version=Public FAQ, legal pages and vendor replies retrieved 2026-10-21; no authenticated deployment observed; plan=Individual Free, Pro and Premier evidence; no organizational plan or entitlement established; add_ons=Optional customer API keys; plan-specific quotas and checkout details not fully public; deployment_mode=not specified; region=US-hosted service with provider processing in US or other countries; no EU/CH-only option established; os=not specified. Limitations: - Documented findings only; unestablished atomic questions remain explicit gaps. - dpa — Is a DPA available, and is it negotiable? [unknown]: Is a DPA available, and is it negotiable? No public negotiable Article28 DPA, explicit revDSG/GDPR processor undertaking, EU AI Act pack or FINMA/sector material was retrieved. The guessed/dpa route did not expose readable content; that is an access gap, not proof of absence. - revdsg_gdpr — Does the DPA cover revDSG and GDPR? [unknown]: Does the DPA cover revDSG and GDPR? No public negotiable Article28 DPA, explicit revDSG/GDPR processor undertaking, EU AI Act pack or FINMA/sector material was retrieved. The guessed/dpa route did not expose readable content; that is an access gap, not proof of absence. - governing_law — Which law and court govern the contract? [supported]: Terms section19 uses California law and individual JAMS arbitration in Santa Clara County, with small-claims/injunction exceptions and state/federal Santa Clara courts where arbitration does not apply. Sources: t3-chat:t3-chat:terms. - ai_act_materials — Is EU AI Act documentation provided? [unknown]: Is EU AI Act documentation provided? No public negotiable Article28 DPA, explicit revDSG/GDPR processor undertaking, EU AI Act pack or FINMA/sector material was retrieved. The guessed/dpa route did not expose readable content; that is an access gap, not proof of absence. - sector_materials — Are FINMA or other sector materials provided? [unknown]: Are FINMA or other sector materials provided? No public negotiable Article28 DPA, explicit revDSG/GDPR processor undertaking, EU AI Act pack or FINMA/sector material was retrieved. The guessed/dpa route did not expose readable content; that is an access gap, not proof of absence. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: t3-chat:t3-chat:terms ### DATA-07: Certifications and attestations Judgment rule: Record each certificate or report separately: ISO 27001, 27017, 27018, 27701, 42001; SOC 2 Type I/II with period; BSI C5; TISAX; national schemes. Record issuer, scope (which product and entity), validity and access route. A cloud provider's certificate does not certify the product. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial ISO and penetration-test documents are advertised behind email access; certificates were not independently verified. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - iso_27001: DPA Annex 3 states ISO/IEC 27001:2022 certificate DE-IS-20250256, Proks Certification GmbH, development/operation of MeinGPT SaaS and custom services, valid to 2028-12-16. The certificate/issuer record was not obtained, so this remains a vendor claim. - c5: The register attributes BSI C5 to Hetzner infrastructure, not to the MeinGPT service. - pentest: A SySS 2025 penetration-test certificate is listed but its result/scope and validity were not retrieved. - Unverified soc2, other_certifications: SOC 2 and other product attestations were not established. No email-gated report was requested. - iso_27001 — ISO 27001 with scope and validity? [partial]: DPA Annex 3 states ISO/IEC 27001:2022 certificate DE-IS-20250256, Proks Certification GmbH, development/operation of MeinGPT SaaS and custom services, valid to 2028-12-16. The certificate/issuer record was not obtained, so this remains a vendor claim. Sources: meingpt:meingpt:dpa, meingpt:meingpt:trust. - soc2 — SOC 2 type, period and covered service? [unknown]: SOC 2 type, period and covered service? Unknown. SOC 2 and other product attestations were not established. No email-gated report was requested. - c5 — BSI C5 attestation? [partial]: The register attributes BSI C5 to Hetzner infrastructure, not to the MeinGPT service. Sources: meingpt:meingpt:subprocessors. - other_certifications — Which other certificates (27701, 42001, TISAX)? [unknown]: Which other certificates (27701, 42001, TISAX)? Unknown. SOC 2 and other product attestations were not established. No email-gated report was requested. - report_access — How can customers obtain reports? [supported]: The trust centre asks for email-based access for ISO and SySS 2025 penetration-test documents. Sources: meingpt:meingpt:trust. - pentest — Are independent penetration test summaries available? [partial]: A SySS 2025 penetration-test certificate is listed but its result/scope and validity were not retrieved. Sources: meingpt:meingpt:trust. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:dpa, meingpt:meingpt:trust, meingpt:meingpt:subprocessors #### T3 Chat Verification status: No public information found (not-published) Support: not-published We found no public information on this. A missing finding is not evidence that the product lacks the capability. ### DATA-08: Encryption and key control Judgment rule: Record encryption at rest and in transit, customer-managed keys (BYOK/HYOK), key location, and private networking (private link, IP allow-lists, VPN). #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial Encryption and logical tenancy are documented; key and private-networking details remain incomplete. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Business/Enterprise hosted workspace; CMK entitlement excluded from verified findings; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - CMK is an access/entitlement gap, not an absent capability. - encryption_at_rest — Is data encrypted at rest? [supported]: Security guide states AES-256 at rest and TLS 1.3; DPA Annex 3 includes encryption at rest/transit. Sources: meingpt:meingpt:security, meingpt:meingpt:dpa. - customer_managed_keys — Can the customer manage or hold the keys? [unknown]: Unknown for Business or Enterprise. The security FAQ says Customer Managed Keys are on request but does not identify an eligible tier, price, key custody or location. No plan-scoped CMK commitment was verified. - private_networking — Is private connectivity or IP restriction available? [unknown]: Is private connectivity or IP restriction available? Unknown. A private-link/IP allowlist/VPN product entitlement was not established; Outpost outbound TLS is a source connector rather than a private SaaS endpoint. - tenant_isolation — How are customers separated in shared SaaS? [partial]: DPA Annex 3 specifies logical tenant separation and network segmentation; no independent isolation test was run. Sources: meingpt:meingpt:dpa. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:security, meingpt:meingpt:dpa #### T3 Chat Verification status: No public information found (not-published) Support: not-published We found no public information on this. A missing finding is not evidence that the product lacks the capability. ### DATA-09: Professional secrecy Judgment rule: For professional secrecy holders (Berufsgeheimnisträger: lawyers, doctors, auditors, tax advisers) under § 203 StGB (DE) and Art. 321 StGB (CH), record whether the vendor signs a contractual secrecy undertaking or a § 203 agreement that binds its staff and its sub-processors; whether this comes as a dedicated offering or a paid add-on, and on which plan; who can access content in plain text (vendor staff, sub-processors, model providers); and the documentation provided. A DPA or a no-training clause alone is not a secrecy undertaking. Banking secrecy and FINMA outsourcing belong to DATA-06, not here. Record contract facts, not a legal assessment. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial The base AGB contains conditional professional-secrecy duties. Eligibility and executed terms of the separately marketed Governance addendum remain unknown. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked2026-10-21; plan=Business/Enterprise customer contracts incorporating AGB section12(4); only the base-terms obligation is assessed; add_ons=Separate Governance addendum excluded from verified entitlement; contract required; deployment_mode=multi-tenant-saas; region=Germany/EU platform; model route and deployment exceptions apply; os=not specified. Limitations: - The base-terms duties and a separately executed Governance addendum are separate legal documents. No claim that one depends on purchase of the other is made. - secrecy_undertaking — Does the vendor sign a secrecy undertaking or § 203 StGB / Art. 321 StGB agreement covering its own staff? [partial]: For contracts incorporating the published AGB, section12(4) applies when processed data fall under professional secrecy, including §203. It imposes indefinite confidentiality, access only as necessary to perform services and acknowledgement of criminal duties. The clause states no Governance prerequisite. It does not establish a separately executed addendum or Swiss Art321 coverage. Sources: meingpt:meingpt:secrecy-contract-rev2. - subprocessor_coverage — Does the undertaking extend to its sub-processors, and is that documented? [partial]: Section12(4) requires staff/subcontractors who can access covered data to be bound in at least text form. Subcontracting and services outside the EU require text consent, deemed granted for known providers at contract start. Provider-specific executed undertakings were not retrieved. Sources: meingpt:meingpt:secrecy-contract-rev2. - secrecy_offering — Is it a dedicated offering or add-on, and on which plan? [unknown]: Unknown whether the separate Governance §203 addendum is available with self-service Business or only a contracted platform offer. The calculator lists it under Governance at EUR1.35/user/month annually, but this is not proof of eligibility or a fee for the base AGB duties. Sources: meingpt:meingpt:prices-rev2. - content_access — Who can access content in plain text: vendor staff, sub-processors, model providers? [partial]: The base AGB restrict knowledge of covered data to what is necessary for the contracted service. The clause does not promise that staff or model providers can never see plaintext. Agreement-specific access restrictions beyond section12(4) were not verified. Sources: meingpt:meingpt:secrecy-contract-rev2. - secrecy_documentation — Is the offering publicly documented (contract template, data-flow description, professional-body guidance)? [partial]: The public AGB section12(4) is the verified contract text. A separately executed/template Governance addendum and provider-by-provider undertakings were not obtained. Sources: meingpt:meingpt:secrecy-contract-rev2. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:secrecy-contract-rev2, meingpt:meingpt:prices-rev2 #### T3 Chat Verification status: Agent cross-checked (agent-cross-checked) Support: partial Public policies describe vendor/provider content processing, but professional-secrecy contracts and a business plan remain unverified. Availability: generally-available Delivery: native Scope: surfaces=T3 Chat browser application; version=Public FAQ, legal pages and vendor replies retrieved 2026-10-21; no authenticated deployment observed; plan=Individual account service; no business/professional-secrecy offer verified; add_ons=Optional customer API keys; plan-specific quotas and checkout details not fully public; deployment_mode=not specified; region=US/international policy; no DE/CH-specific arrangement verified; os=not specified. Limitations: - No-training statements and personal account terms are not a professional-secrecy undertaking. - No legal suitability or compliance conclusion. - secrecy_undertaking — Does the vendor sign a secrecy undertaking or § 203 StGB / Art. 321 StGB agreement covering its own staff? [unknown]: No public DPA or signed/signon §203 StGB/Art321 StGB professional-secrecy undertaking covering T3 staff was retrieved from the Terms, privacy/security policies or targeted first-party searches. - subprocessor_coverage — Does the undertaking extend to its sub-processors, and is that documented? [unknown]: The privacy policy names hosting/model services but does not establish a dedicated professional-secrecy undertaking binding them. Its existence and flow-down are unknown. - secrecy_offering — Is it a dedicated offering or add-on, and on which plan? [unknown]: A team/business offer itself remains unverified; no professional-secrecy-specific tier, dedicated package or paid add-on was established. - content_access — Who can access content in plain text: vendor staff, sub-processors, model providers? [partial]: Privacy policy sends input to selected OpenAI/Google/Anthropic/OpenRouter services; Terms reserve vendor review/removal of content upon apparent violations. This establishes processing/review possibilities, not an exhaustive plaintext staff/provider access matrix or protected-profession arrangement. Sources: t3-chat:t3-chat:privacy, t3-chat:t3-chat:terms. - secrecy_documentation — Is the offering publicly documented (contract template, data-flow description, professional-body guidance)? [unknown]: Public consumer/service Terms and privacy/security pages were reviewed, with targeted §203/Art321/professional-secrecy searches. No specific template, protected-profession data-flow statement or professional-body guidance was retrieved; no private agreement was requested. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: t3-chat:t3-chat:privacy, t3-chat:t3-chat:terms ## Commercials and vendor ### VEN-01: Pricing model and list prices Judgment rule: Record pricing unit (per seat, per active user, usage-based, platform fee, hybrid), list prices per plan in original currency with date, included usage and credits, add-on prices (connectors, agents, image generation) and overage rates. Cite public prices or dated quotes. Per-seat versus usage is a fact, not a quality judgment. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial Self-service Business costs EUR29/user/month excluding VAT and includes standard-model use. Enterprise and platform-module pricing are outside this scoped offer. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Self-service Business, monthly billing only; contracted Enterprise/platform calculator excluded; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - The progressive platform calculator is not used as a Business tariff. Its contract eligibility remains outside this finding. - pricing_unit — What is the pricing unit? [supported]: Business is sold per user per month through self-service Stripe checkout. Premium/reasoning models, API use and meeting-bot recordings are separately metered. Sources: meingpt:meingpt:billing-rev2. - list_prices — What are list prices per plan, with date and currency? [supported]: Self-service Business is EUR29/user/month, excluding VAT, on monthly billing. No annual discount or progressive marginal tiers are applied to this offer. Sources: meingpt:meingpt:billing-rev2. - included_usage — What usage is included? [partial]: Standard-model use in Business is unlimited according to the billing guide. Premium/reasoning models, API and meeting bots remain additional; the exact standard-model catalog and any applicable workload restrictions are not fixed by this page. Sources: meingpt:meingpt:billing-rev2. - add_ons — Which features cost extra? [unknown]: Unknown for self-service Business. The separate platform calculator has Governance/Outpost modules, but their Business eligibility and package prices were not established. - overage — What do usage beyond limits and premium models cost? [partial]: Business standard models do not consume additional billed usage under the billing guide. Premium/reasoning, API and meeting-bot use draw on workspace credit; meeting bots cost EUR0.80/audio hour. Other route-specific charges depend on the selected service. Sources: meingpt:meingpt:billing-rev2. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:billing-rev2 #### T3 Chat Verification status: Agent cross-checked (agent-cross-checked) Support: partial Individual subscription prices are supported by vendor replies; quotas are cost-based and checkout/organization details are incomplete. Availability: generally-available Delivery: native Scope: surfaces=T3 Chat browser application; version=Public FAQ, legal pages and vendor replies retrieved 2026-10-21; no authenticated deployment observed; plan=Individual Free, Pro and Premier evidence; no organizational plan or entitlement established; add_ons=Optional customer API keys; plan-specific quotas and checkout details not fully public; deployment_mode=not specified; region=US-hosted service with provider processing in US or other countries; no EU/CH-only option established; os=not specified. Limitations: - Staff pricing replies are first-party but undated; current checkout amounts and tax/plan availability were not observed. - pricing_unit — What is the pricing unit? [supported]: Evidence describes monthly individual Pro/Premier subscriptions, optional customer provider keys and cost-based included use. No organizational seat/pool contract is established. Sources: t3-chat:t3-chat:billing-change, t3-chat:t3-chat:premier, t3-chat:t3-chat:faq-usage. - list_prices — What are list prices per plan, with date and currency? [partial]: Vendor replies confirm USD8/month paid-tier context and Premier USD50/month plus applicable tax. No checkout was completed and no business price was retrieved; dates of the undated replies are unknown. Sources: t3-chat:t3-chat:billing-change, t3-chat:t3-chat:premier, t3-chat:t3-chat:price-visibility. - included_usage — What usage is included? [supported]: Included usage is represented by Base and Overage balances, refilling four-hourly/monthly. The FAQ’s20/200 example is not the actual plan quota and the former1500+100 message claim is not used. Sources: t3-chat:t3-chat:faq-usage, t3-chat:t3-chat:billing-change. - add_ons — Which features cost extra? [partial]: Customer OpenRouter/OpenAI API keys incur direct provider charges. Staff says a top-up product has not been offered since the new billing model and options are only being considered. Sources: t3-chat:t3-chat:premier, t3-chat:t3-chat:overage, t3-chat:t3-chat:byok-openai. - overage — What do usage beyond limits and premium models cost? [partial]: The Overage bucket is a monthly included-usage bucket, not an established published dollar-per-token overage tariff. Extra usage via own keys is billed by the customer provider; current model/route tariffs are not fully mapped. Sources: t3-chat:t3-chat:faq-usage, t3-chat:t3-chat:premier. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: t3-chat:t3-chat:billing-change, t3-chat:t3-chat:premier, t3-chat:t3-chat:faq-usage, t3-chat:t3-chat:price-visibility, t3-chat:t3-chat:overage, t3-chat:t3-chat:byok-openai ### VEN-02: Scenario cost Judgment rule: Price three workloads for 12 months: 200, 1,000 and 5,000 users with a disclosed light/regular/power usage mix, in CHF and EUR excluding VAT. Include licences, platform fees, add-ons, model usage and, for self-hosted options, a stated operating assumption. Separate fixed, variable, included and unknown costs. No invented discounts. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial Business-only seat floors are EUR69600 / EUR348000 / EUR1740000 for200/1000/5000 users over12months, each plus unknown workload and contract components. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Self-service Business at EUR29/user/month, paid monthly for12months; Business standard-chat floor only; add_ons=No paid add-ons included; required extras unknown; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - ALM-14 shared cost assumptions v1.4, 2026-10-21:12months, list prices, excluding VAT;60% light100,30% regular400,10% power1500 messages/user/month;3000 input and600 output tokens/message. Weighted330messages,0.99M input/0.198M output per user/month. Token mix is used only on separately billed routes; this Business standard-chat floor has included usage. - ECB2026-10-20 EUR1=USD1.1215=CHF0.9237. Vendor currency EUR is primary. Contract-specific extras remain unknown. - cost_200 — Year-one cost at 200 users? [partial]: ALM-14 shared cost assumptions v1.4, 2026-10-21. Self-service Business seat floor for 200 users over12months: from EUR69600.00 / CHF64289.52 + unknown, excluding VAT. Uses EUR29/user/month with monthly billing and no invented annual discount. Standard-model chat is included, so no separate token bill is added for that included route. Eligibility at this size, the exact standard-model catalog needed for the workload, premium/API/meeting use, add-ons and Enterprise services remain unknown. This is a Business floor, not an Enterprise quote. Sources: meingpt:meingpt:billing-rev2, meingpt:meingpt:ecb-fx. - cost_1000 — Year-one cost at 1,000 users? [partial]: ALM-14 shared cost assumptions v1.4, 2026-10-21. Self-service Business seat floor for 1000 users over12months: from EUR348000.00 / CHF321447.60 + unknown, excluding VAT. Uses EUR29/user/month with monthly billing and no invented annual discount. Standard-model chat is included, so no separate token bill is added for that included route. Eligibility at this size, the exact standard-model catalog needed for the workload, premium/API/meeting use, add-ons and Enterprise services remain unknown. This is a Business floor, not an Enterprise quote. Sources: meingpt:meingpt:billing-rev2, meingpt:meingpt:ecb-fx. - cost_5000 — Year-one cost at 5,000 users? [partial]: ALM-14 shared cost assumptions v1.4, 2026-10-21. Self-service Business seat floor for 5000 users over12months: from EUR1740000.00 / CHF1607238.00 + unknown, excluding VAT. Uses EUR29/user/month with monthly billing and no invented annual discount. Standard-model chat is included, so no separate token bill is added for that included route. Eligibility at this size, the exact standard-model catalog needed for the workload, premium/API/meeting use, add-ons and Enterprise services remain unknown. This is a Business floor, not an Enterprise quote. Sources: meingpt:meingpt:billing-rev2, meingpt:meingpt:ecb-fx. - variable_share — Which part depends on usage? [partial]: Business standard-model chat is included. A model-preserving workload outside that standard catalog, premium/reasoning models, API and meeting-bot charges cannot be quantified without the actual route and contract; the variable remainder stays unknown. No provider proxy or progressive-platform tariff is substituted. Sources: meingpt:meingpt:billing-rev2. - unknown_parts — Which costs could not be priced? [partial]: Business eligibility for200/1000/5000 users, standard-model mapping, premium/tool use, required Governance/Outpost, rollout and Enterprise services remain unknown. No progressive-platform amount is mixed into the Business floor. Sources: meingpt:meingpt:billing-rev2. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:billing-rev2, meingpt:meingpt:ecb-fx #### T3 Chat Verification status: Agent cross-checked (agent-cross-checked) Support: partial Multiplying individual list prices yields conditional licence floors, not a verified organization quote or usage-complete TCO. Availability: generally-available Delivery: native Scope: surfaces=T3 Chat browser application; version=Public FAQ, legal pages and vendor replies retrieved 2026-10-21; no authenticated deployment observed; plan=Conditional independent Pro/Premier account multiplication; no organization plan verified; add_ons=Optional customer API keys; plan-specific quotas and checkout details not fully public; deployment_mode=not specified; region=US-hosted service with provider processing in US or other countries; no EU/CH-only option established; os=not specified. Limitations: - Known price floor + unknown components; no complete year-one TCO, seat commitment or sufficient-usage guarantee. - cost_200 — Year-one cost at 200 users? [partial]: ALM-14 shared cost assumptions v1.4,2026-10-21:12 months,list prices,exVAT,no negotiated discounts;60% light100/30% regular400/10% power1500 messages/user/month =>330,each3000 input+600 output tokens incl retrieval. Vendor subscription usage remains subject to cost/refill limits; no fixed default/recommended model or complete tariff is published, so coverage and variable extra cost stay unknown. ECB2026-10-20 EUR1=USD1.1215=CHF0.9237; divideUSD by1.1215 then multiplyEUR by0.9237. For200 separate accounts: Pro from USD19,200/EUR17,119.93/CHF15,813.68 + unknown; Premier reference USD120,000/EUR106,999.55/CHF98,835.49 + unknown. Bulk purchase and business-governance fit unverified. Sources: t3-chat:t3-chat:billing-change, t3-chat:t3-chat:premier, t3-chat:t3-chat:faq-usage, t3-chat:t3-chat:fx. - cost_1000 — Year-one cost at 1,000 users? [partial]: Same assumptions,1000 separate accounts: Pro from USD96,000/EUR85,599.64/CHF79,068.39 + unknown; Premier reference USD600,000/EUR534,997.77/CHF494,177.44 + unknown. Not a1000-seat quote. Sources: t3-chat:t3-chat:billing-change, t3-chat:t3-chat:premier, t3-chat:t3-chat:faq-usage, t3-chat:t3-chat:fx. - cost_5000 — Year-one cost at 5,000 users? [partial]: Same assumptions,5000 separate accounts: Pro from USD480,000/EUR427,998.22/CHF395,341.95 + unknown; Premier reference USD3,000,000/EUR2,674,988.85/CHF2,470,887.20 + unknown. Not a5000-seat quote. Sources: t3-chat:t3-chat:billing-change, t3-chat:t3-chat:premier, t3-chat:t3-chat:faq-usage, t3-chat:t3-chat:fx. - variable_share — Which part depends on usage? [partial]: Included usage consumes a cost meter with no complete published tariff/default model. For an optional direct-provider BYOK comparison only, the shared gpt-6-luna effort-none Standard proxy is USD0.10 input/USD0.50 output per1M and USD2.376/user/year at the shared mix, excluding cache/tools/embeddings/regional uplift. This exact model/route is not verified selectable in T3 and is not added to the subscription floor as though it were a vendor tariff. Sources: t3-chat:t3-chat:faq-usage, t3-chat:t3-chat:openai-price, t3-chat:t3-chat:openai-model. - unknown_parts — Which costs could not be priced? [supported]: Unpriced: enterprise/bulk entitlement, organization features, exact current checkout prices, model availability and complete token/search/file tariff, whether the shared workload fits refill quotas, direct-provider/OpenRouter route charges, regional processing and support/implementation. No assumed wholesale discount or model-cost subsidy. Sources: t3-chat:t3-chat:premier, t3-chat:t3-chat:price-visibility, t3-chat:t3-chat:faq-usage, t3-chat:t3-chat:byok-router. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: t3-chat:t3-chat:billing-change, t3-chat:t3-chat:premier, t3-chat:t3-chat:faq-usage, t3-chat:t3-chat:fx, t3-chat:t3-chat:openai-price, t3-chat:t3-chat:openai-model, t3-chat:t3-chat:price-visibility, t3-chat:t3-chat:byok-router ### VEN-03: Evaluation access Judgment rule: Distinguish free tier, self-serve trial, sales-led pilot, paid proof of concept and open-source self-install. Record card requirement, duration, seat limit, and which features (SSO, connectors, admin) the trial includes. Map which shared scenarios can be run without sales approval. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial A self-serve 10-day trial exists, while several enterprise features still need enablement. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - free_tier: Trial expiry locks the workspace even for free models; the guide does not describe a permanent free business workspace. - trial_scope: Most base functions are included; personal API keys are off, and request-only meeting/live recording and automations are initially off as in paid plans. SCIM/SSO enterprise enablement is not guaranteed. - sales_gate: Base trial is self-serve; request-enabled features and Enterprise terms are not self-service. No sales contact was made. - free_tier — Is there a free tier for business use? [partial]: Trial expiry locks the workspace even for free models; the guide does not describe a permanent free business workspace. Sources: meingpt:meingpt:billing. - self_serve_trial — Is a self-serve business trial available? [supported]: 10 days, EUR 5 after phone verification, no payment details or automatic paid conversion, up to ten emailed invitees. Sources: meingpt:meingpt:billing. - trial_scope — Do trials include SSO, connectors and admin features? [partial]: Most base functions are included; personal API keys are off, and request-only meeting/live recording and automations are initially off as in paid plans. SCIM/SSO enterprise enablement is not guaranteed. Sources: meingpt:meingpt:billing. - sales_gate — Does meaningful evaluation require sales contact? [partial]: Base trial is self-serve; request-enabled features and Enterprise terms are not self-service. No sales contact was made. Sources: meingpt:meingpt:billing. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:billing #### T3 Chat Verification status: Agent cross-checked (agent-cross-checked) Support: partial Free usage is in the Terms; a business-specific trial and enterprise evaluation path remain unverified. Availability: generally-available Delivery: native Scope: surfaces=T3 Chat browser application; version=Public FAQ, legal pages and vendor replies retrieved 2026-10-21; no authenticated deployment observed; plan=Individual Free, Pro and Premier evidence; no organizational plan or entitlement established; add_ons=Optional customer API keys; plan-specific quotas and checkout details not fully public; deployment_mode=not specified; region=US-hosted service with provider processing in US or other countries; no EU/CH-only option established; os=not specified. Limitations: - Documented findings only; unestablished atomic questions remain explicit gaps. - free_tier — Is there a free tier for business use? [partial]: Terms say services are free up to unspecified usage limits and paid resources may be required. Organizational acceptance is allowed, but no free business admin entitlement is established. Sources: t3-chat:t3-chat:terms. - self_serve_trial — Is a self-serve business trial available? [unknown]: Is a self-serve business trial available? No timed self-serve business trial, SSO/connector/admin trial entitlement or enterprise evaluation agreement was retrieved. No account was created. - trial_scope — Do trials include SSO, connectors and admin features? [unknown]: Do trials include SSO, connectors and admin features? No timed self-serve business trial, SSO/connector/admin trial entitlement or enterprise evaluation agreement was retrieved. No account was created. - sales_gate — Does meaningful evaluation require sales contact? [partial]: A public browser service and free-limited offer are documented; business-specific meaningful evaluation and any sales gate cannot be determined without an established business offer. Sources: t3-chat:t3-chat:terms. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: t3-chat:t3-chat:terms ### VEN-04: Contract terms and pricing transparency Judgment rule: Record minimum seats or commitment, contract length, monthly versus annual billing, price-change rules, cancellation and downgrade terms, and invoicing in CHF/EUR. Classify pricing as transparent, partially specified, quote-only or unknown, naming the missing fields. Quote-only pricing is a disclosed constraint. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial Public price fields and German contract defaults exist; customer-specific terms and CHF invoicing remain open. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - minimum_commitment: Public calculator starts at one user; individually negotiated Enterprise or on-prem minimums are not stated. - local_invoicing: SelectCode GmbH is the German contracting entity and prices are EUR; CHF invoices were not confirmed. - transparency_class: Partially specified: public marginal seat/model/add-on rates, but PII base, services and deployments are quote-only and Business packaging differs. - minimum_commitment — What seat or spend minimum applies? [partial]: Public calculator starts at one user; individually negotiated Enterprise or on-prem minimums are not stated. Sources: meingpt:meingpt:prices. - contract_length — What contract lengths are offered? [supported]: Public calculator has annual/monthly options; terms default to one year absent another agreement. Sources: meingpt:meingpt:prices, meingpt:meingpt:terms. - cancellation — How can the contract be cancelled or reduced? [supported]: Absent other terms, three months before term-end, text/email or admin cancellation; downgrades/user reductions effective following month, no current-month refund. Price changes require the stated notice/consent procedure. Sources: meingpt:meingpt:terms. - local_invoicing — Can it be invoiced in CHF or EUR from a European entity? [partial]: SelectCode GmbH is the German contracting entity and prices are EUR; CHF invoices were not confirmed. Sources: meingpt:meingpt:terms. - transparency_class — Is pricing transparent, partial, quote-only or unknown? [partial]: Partially specified: public marginal seat/model/add-on rates, but PII base, services and deployments are quote-only and Business packaging differs. Sources: meingpt:meingpt:prices, meingpt:meingpt:billing. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:prices, meingpt:meingpt:terms, meingpt:meingpt:billing #### T3 Chat Verification status: Agent cross-checked (agent-cross-checked) Support: partial Individual monthly upgrades/downgrades are documented; business minimums and European invoicing remain unverified. Availability: generally-available Delivery: native Scope: surfaces=T3 Chat browser application; version=Public FAQ, legal pages and vendor replies retrieved 2026-10-21; no authenticated deployment observed; plan=Individual Free, Pro and Premier evidence; no organizational plan or entitlement established; add_ons=Optional customer API keys; plan-specific quotas and checkout details not fully public; deployment_mode=not specified; region=US-hosted service with provider processing in US or other countries; no EU/CH-only option established; os=not specified. Limitations: - Documented findings only; unestablished atomic questions remain explicit gaps. - minimum_commitment — What seat or spend minimum applies? [partial]: The public evidence concerns individual accounts and monthly prices, not a minimum team-seat/spend commitment. Sources: t3-chat:t3-chat:terms, t3-chat:t3-chat:premier. - contract_length — What contract lengths are offered? [supported]: Vendor replies confirm monthly Premier pricing and billing periods; no signed annual or business-term offer was retrieved. Sources: t3-chat:t3-chat:premier. - cancellation — How can the contract be cancelled or reduced? [supported]: Terms permit account cancellation at any time without refunds. Staff says upgrades take effect immediately and downgrades at the billing-period end. Sources: t3-chat:t3-chat:terms, t3-chat:t3-chat:premier. - local_invoicing — Can it be invoiced in CHF or EUR from a European entity? [unknown]: Can it be invoiced in CHF or EUR from a European entity? No team minimum, European contracting entity or CHF/EUR invoice entitlement was retrieved. - transparency_class — Is pricing transparent, partial, quote-only or unknown? [supported]: Partial: some individual prices are confirmed in undated staff replies, complete final checkout requires billing information, quotas are not a public token tariff, and a business offer remains unverified. Sources: t3-chat:t3-chat:price-visibility, t3-chat:t3-chat:premier, t3-chat:t3-chat:faq-usage. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: t3-chat:t3-chat:terms, t3-chat:t3-chat:premier, t3-chat:t3-chat:price-visibility, t3-chat:t3-chat:faq-usage ### VEN-05: Source licence Judgment rule: Record licence per component (client, server, connectors, hosted extensions) and revision. Distinguish OSI open source, source-available with restrictions (branding clauses, user caps), and closed. Repository visibility is not a licence classification. Licences change between releases: record the current terms, and never carry a fork's historical licence over to the upstream product or the reverse. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial The public AGB establish a commercial subscription/rental use licence with contract-limited rights. Public component source licences and a distinct enterprise source-edition mapping remain unverified. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - Contractual service-use rights do not establish the licence of every server/client/connector dependency. - No public implementation repository or enterprise source-edition map was identified. - licence_type — Which licence applies to each component? [partial]: SelectCode AGB describe the software as a subscription/rental and grant a non-exclusive use right limited to the contract term. Public display, adaptation, translation and decompilation of the software are restricted subject to statutory rights. This establishes the commercial service-use licence; a client/server/connector source-licence inventory and current implementation revision remain unpublished. Sources: meingpt:meingpt:terms-20261026. - branding_restrictions — Do licence terms restrict branding or user counts? [partial]: AGB require purchased licences for majority-owned affiliates and give no independent right to sublicense or transfer usage rights. They restrict adapting the software. A numerical user cap or a component-specific branding licence is not established by these clauses. Sources: meingpt:meingpt:terms-20261026. - enterprise_features_closed — Are enterprise features in a separate closed edition? [unknown]: Are enterprise features in a separate closed edition? Unknown. The public commercial terms describe rented software, but no public client/server/connector source licence or current repository revision was located. Repository visibility was not treated as a licence. Rechecked 2026-10-26 in Allgemeine Geschäftsbedingungen | MeinGPT | KI-Use-Case-Plattform für den Mittelstand. Official links and targeted first-party searches yielded no further plan-scoped specification for the remaining atom. Checked 2026-10-26; agent cross-checked 2026-10-26; check again by 2026-12-25. Sources: meingpt:meingpt:terms-20261026 #### T3 Chat Verification status: Agent cross-checked (agent-cross-checked) Support: partial The hosted service is proprietary; clone hackathons and T3 Code are separate products. Availability: generally-available Delivery: native Scope: surfaces=T3 Chat browser application; version=Public FAQ, legal pages and vendor replies retrieved 2026-10-21; no authenticated deployment observed; plan=Individual Free, Pro and Premier evidence; no organizational plan or entitlement established; add_ons=Optional customer API keys; plan-specific quotas and checkout details not fully public; deployment_mode=not specified; region=US-hosted service with provider processing in US or other countries; no EU/CH-only option established; os=not specified. Limitations: - Documented findings only; unestablished atomic questions remain explicit gaps. - licence_type — Which licence applies to each component? [supported]: Terms grant a personal nonassignable service licence and prohibit copying, modification, reverse engineering and source extraction, subject to third-party software rights. Sources: t3-chat:t3-chat:terms. - branding_restrictions — Do licence terms restrict branding or user counts? [supported]: Terms reserve vendor trademarks and prohibit sublicensing/renting/providing service access to others; each account is personal and cannot be shared. Sources: t3-chat:t3-chat:terms. - enterprise_features_closed — Are enterprise features in a separate closed edition? [unknown]: Are enterprise features in a separate closed edition? No supported self-host licence, open-source core/enterprise split or enterprise edition licence was retrieved. Public third-party clones are not licensed distributions of the hosted product. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: t3-chat:t3-chat:terms ### VEN-06: Data export and exit Judgment rule: Record admin export of chats, assistants, knowledge bases, prompts and audit logs, formats, export completeness, user self-export, and what survives contract end. Run an admin export on a trial. Exported chats alone do not make assistants and knowledge portable. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial Individual downloads exist but a bulk chat export is explicitly unavailable. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - chat_export: Security FAQ explicitly says there is currently no function to export all chats at once. - export_format: Individual artifacts/slides/files/images can be downloaded; administrative logs have CSV/JSON exports. This does not establish portable assistants or a complete knowledge export. - post_termination_window: Terms permit export until termination takes effect; later access is blocked. DPA allows an individually agreed export window and return/deletion within 30 days after it. - Unverified assistant_export, knowledge_export: No reusable assistant/prompt or knowledge-bundle export specification was located and no admin export fixture was run. - chat_export — Can admins export all chats? [unsupported]: Security FAQ explicitly says there is currently no function to export all chats at once. Sources: meingpt:meingpt:security. - assistant_export — Can assistants and prompts be exported in a reusable format? [unknown]: Can assistants and prompts be exported in a reusable format? Unknown. No reusable assistant/prompt or knowledge-bundle export specification was located and no admin export fixture was run. - knowledge_export — Can knowledge bases and their sources be exported? [unknown]: Can knowledge bases and their sources be exported? Unknown. No reusable assistant/prompt or knowledge-bundle export specification was located and no admin export fixture was run. - export_format — Are formats documented and machine-readable? [partial]: Individual artifacts/slides/files/images can be downloaded; administrative logs have CSV/JSON exports. This does not establish portable assistants or a complete knowledge export. Sources: meingpt:meingpt:security, meingpt:meingpt:audit-export. - post_termination_window — How long can data be retrieved after termination? [partial]: Terms permit export until termination takes effect; later access is blocked. DPA allows an individually agreed export window and return/deletion within 30 days after it. Sources: meingpt:meingpt:terms, meingpt:meingpt:dpa. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:security, meingpt:meingpt:audit-export, meingpt:meingpt:terms, meingpt:meingpt:dpa #### T3 Chat Verification status: Agent cross-checked (agent-cross-checked) Support: partial Terms require pre-termination export, but a documented organizational export package was not established. Availability: generally-available Delivery: native Scope: surfaces=T3 Chat browser application; version=Public FAQ, legal pages and vendor replies retrieved 2026-10-21; no authenticated deployment observed; plan=Individual Free, Pro and Premier evidence; no organizational plan or entitlement established; add_ons=Optional customer API keys; plan-specific quotas and checkout details not fully public; deployment_mode=not specified; region=US-hosted service with provider processing in US or other countries; no EU/CH-only option established; os=not specified. Limitations: - Documented findings only; unestablished atomic questions remain explicit gaps. - chat_export — Can admins export all chats? [unknown]: Can admins export all chats? No first-party complete-admin-chat export procedure/format, reusable assistant/prompt export or KB/source export was retrieved. Community migration scripts and export comments are leads, not supported portability guarantees. - assistant_export — Can assistants and prompts be exported in a reusable format? [unknown]: Can assistants and prompts be exported in a reusable format? No first-party complete-admin-chat export procedure/format, reusable assistant/prompt export or KB/source export was retrieved. Community migration scripts and export comments are leads, not supported portability guarantees. - knowledge_export — Can knowledge bases and their sources be exported? [unknown]: Can knowledge bases and their sources be exported? No first-party complete-admin-chat export procedure/format, reusable assistant/prompt export or KB/source export was retrieved. Community migration scripts and export comments are leads, not supported portability guarantees. - export_format — Are formats documented and machine-readable? [unknown]: Are formats documented and machine-readable? No first-party complete-admin-chat export procedure/format, reusable assistant/prompt export or KB/source export was retrieved. Community migration scripts and export comments are leads, not supported portability guarantees. - post_termination_window — How long can data be retrieved after termination? [partial]: Users must export before termination. If the vendor terminates for convenience it will endeavor to allow a reasonable retrieval opportunity; no fixed post-termination window is guaranteed. Sources: t3-chat:t3-chat:terms. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: t3-chat:t3-chat:terms ### VEN-07: Support and service levels Judgment rule: Record support channels, languages (German, French, Italian), hours and time zone, local or regional presence, documented response times, uptime SLA and credits, public status page and incident history. Popularity is not a reliability score. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial German vendor support and a public incident history are documented; an uptime credit schedule was not found. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - support_channels: Terms provide email support Monday-Friday 09:00-17:00; timezone and plan-specific guaranteed response times are not stated. - Unverified support_languages, uptime_sla: Contractual German/French/Italian support coverage, uptime commitments and credits were not established in public terms; German documentation alone does not prove language coverage. - support_channels — Which support channels are available on each plan? [partial]: Terms provide email support Monday-Friday 09:00-17:00; timezone and plan-specific guaranteed response times are not stated. Sources: meingpt:meingpt:terms. - support_languages — Is support available in German, French or Italian? [unknown]: Is support available in German, French or Italian? Unknown. Contractual German/French/Italian support coverage, uptime commitments and credits were not established in public terms; German documentation alone does not prove language coverage. - regional_presence — Is there a DACH or Swiss entity, team or partner? [supported]: SelectCode GmbH contracts from Unterhaching, Germany. Sources: meingpt:meingpt:terms. - uptime_sla — Is there an uptime SLA with credits? [unknown]: Is there an uptime SLA with credits? Unknown. Contractual German/French/Italian support coverage, uptime commitments and credits were not established in public terms; German documentation alone does not prove language coverage. - status_page — Is there a public status page with history? [supported]: Public status.meingpt.com lists the platform and historical incidents; uptime is not an SLA. Sources: meingpt:meingpt:status. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:terms, meingpt:meingpt:status #### T3 Chat Verification status: Agent cross-checked (agent-cross-checked) Support: partial Email support and official feedback are public, without a retrieved SLA or regional-language contract. Availability: generally-available Delivery: native Scope: surfaces=T3 Chat browser application; version=Public FAQ, legal pages and vendor replies retrieved 2026-10-21; no authenticated deployment observed; plan=Individual Free, Pro and Premier evidence; no organizational plan or entitlement established; add_ons=Optional customer API keys; plan-specific quotas and checkout details not fully public; deployment_mode=not specified; region=US-hosted service with provider processing in US or other countries; no EU/CH-only option established; os=not specified. Limitations: - Documented findings only; unestablished atomic questions remain explicit gaps. - support_channels — Which support channels are available on each plan? [supported]: The FAQ lists a support email and official feature/bug board; plan-specific response targets are not published there. Sources: t3-chat:t3-chat:faq-support. - support_languages — Is support available in German, French or Italian? [unknown]: Is support available in German, French or Italian? No German/French/Italian support commitment, credit-bearing uptime SLA or verified public status/history service was retrieved. Guessed status.t3.chat failed certificate validation; no incident conclusion follows. - regional_presence — Is there a DACH or Swiss entity, team or partner? [partial]: The vendor publishes a San Francisco contact address. A DACH/Swiss entity, partner or support team was not established. Sources: t3-chat:t3-chat:privacy. - uptime_sla — Is there an uptime SLA with credits? [unknown]: Is there an uptime SLA with credits? No German/French/Italian support commitment, credit-bearing uptime SLA or verified public status/history service was retrieved. Guessed status.t3.chat failed certificate validation; no incident conclusion follows. - status_page — Is there a public status page with history? [unknown]: Is there a public status page with history? No German/French/Italian support commitment, credit-bearing uptime SLA or verified public status/history service was retrieved. Guessed status.t3.chat failed certificate validation; no incident conclusion follows. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: t3-chat:t3-chat:faq-support, t3-chat:t3-chat:privacy ### VEN-08: Onboarding, training and adoption Judgment rule: Record vendor-led onboarding, admin training, end-user training (live, e-learning, languages), use-case libraries, champion programmes, change-management materials and their cost. Distinguish included services from paid professional services and partner offerings. #### meinGPT Verification status: Agent cross-checked (agent-cross-checked) Support: partial Learning paths and business examples are built in; personal rollout services are quote-based. Availability: generally-available Delivery: native Scope: surfaces=meinGPT hosted workspace; version=Public documentation checked 2026-10-21; plan=Platform licence; Governance, Enterprise API and request-only features identified below; add_ons=none stated; deployment_mode=multi-tenant-saas; region=Germany platform storage; model processing depends on provider, model and agreed DPA; os=not specified. Limitations: - onboarding_service: Personal guidance covers selection, pilot and rollout on an individual offer; universal included consulting is not promised. - user_training: The learning area has guided practical courses; Governance enables mandatory assignments. Complete course language coverage was not stated. - adoption_programme: Pilot/adoption workspaces typically run a three-month fixed-price pilot with workshops; a defined champion programme was not found. - onboarding_service — Is guided onboarding included? [partial]: Personal guidance covers selection, pilot and rollout on an individual offer; universal included consulting is not promised. Sources: meingpt:meingpt:prices. - user_training — Is end-user training offered, in which languages? [partial]: The learning area has guided practical courses; Governance enables mandatory assignments. Complete course language coverage was not stated. Sources: meingpt:meingpt:academy, meingpt:meingpt:audit. - use_case_library — Are role-based use cases and templates provided? [supported]: Role/department personalization and business-use-case material are documented. Sources: meingpt:meingpt:academy. - adoption_programme — Is there a champion or adoption programme? [partial]: Pilot/adoption workspaces typically run a three-month fixed-price pilot with workshops; a defined champion programme was not found. Sources: meingpt:meingpt:billing. - service_cost — Which services cost extra? [supported]: Personal guidance is quote-only; terms bill services by time/person-days or individually agreed fixed price. Sources: meingpt:meingpt:prices, meingpt:meingpt:terms. Checked 2026-10-21; agent cross-checked 2026-10-21; check again by 2026-12-20. Sources: meingpt:meingpt:prices, meingpt:meingpt:academy, meingpt:meingpt:audit, meingpt:meingpt:billing, meingpt:meingpt:terms #### T3 Chat Verification status: No public information found (not-published) Support: not-published We found no public information on this. A missing finding is not evidence that the product lacks the capability. ## Sources ### meingpt:meingpt:rates Alle Modelle | MeinGPT Docs. Public EUR token and service rates include GPT-5 mini at EUR 0.30 input and EUR 2.40 output per million tokens. - URL: https://meingpt.com/docs/admin-guide/pricing - Locator: Text-generation, translation, images and other-functions tables - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:models-rev2 Modelle | MeinGPT Docs. Current documentation lists selectable non-Meta model families, per-chat selection and context-length routing; the lifecycle table contains no current selectable Meta model. - URL: https://meingpt.com/docs/platform/models - Locator: Model picker, model tables and own-LLM-key FAQ - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:lifecycle-rev2 Modell-Lifecycle | MeinGPT Docs. Active models, future retirement dates and automatic replacement rules are published. - URL: https://meingpt.com/docs/admin-guide/model-lifecycle - Locator: Phases, notice period and current-model table - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### t3-chat:t3-chat:faq-training T3 Chat FAQ. The vendor says it does not train models, configures OpenAI/Anthropic/paid Google/OpenRouter no-training where possible, labels exceptions and makes BYOK customers responsible for settings. - URL: https://t3.chat/faq - Locator: Do you use my data for training? - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### t3-chat:t3-chat:billing-change Usage model change. Vendor replies confirm current cost-based usage and that an USD8 subscription can still use expensive models within cost limits. - URL: https://feedback.t3.chat/p/t3-feels-nerfed-now - Locator: Mark Florkowski Team replies - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### t3-chat:t3-chat:r2-paid-tier-controls Pro/Premier vendor replies. The vendor discusses USD8 Pro usage/search costs, optional OpenRouter keys and Premier USD50/month; this does not establish Free parity or a complete model matrix. - URL: https://feedback.t3.chat/p/new-tier-in-between-pro-and-premier - Locator: Mark Florkowski Team: cost explanation, OpenRouter direct billing, USD50 Premier and plan changes - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### t3-chat:t3-chat:r2-pro-context-byok Context and own-key controls. Vendor replies document context restrictions on the USD8 tier and optional own-key fallback without publishing a complete current model/file matrix. - URL: https://feedback.t3.chat/p/premium-requests-for-large-context-windows - Locator: Mark Florkowski Team replies on USD8 limits, higher tier, model switching and OpenRouter fallback - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:lifecycle Modell-Lifecycle | MeinGPT Docs. Active models, future retirement dates and automatic replacement rules are published. - URL: https://meingpt.com/docs/admin-guide/model-lifecycle - Locator: Phases, notice period and current-model table - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:changes Was wir zuletzt released haben | MeinGPT | KI-Use-Case-Plattform für den Mittelstand. Release 2.67.0 on 18 October adds Gemini 4 Argon, retires named models and adds configurable deletion periods. - URL: https://meingpt.com/changelog - Locator: v2.67.0, model retirement and deletion periods - Evidence kind: documented - Source date: 2026-10-18 - Retrieved: 2026-10-21 ### t3-chat:t3-chat:changelog Changelog of new features. A vendor reply says major-feature changelog is in Discord and a website version is intended; the website request remains Planned. - URL: https://feedback.t3.chat/p/changelog-of-new-features - Locator: Planned status and Mark Florkowski Team reply - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:models Modelle | MeinGPT Docs. Users choose models per chat; automatic selection is based on context length; customer LLM keys and endpoints cannot be added. - URL: https://meingpt.com/docs/platform/models - Locator: Model picker, model tables and own-LLM-key FAQ - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:translation Übersetzer | MeinGPT Docs. Default text translation uses Google Cloud Translate and document translation uses DeepL through EU endpoints; glossaries and layout preservation are documented. - URL: https://meingpt.com/docs/platform/translation - Locator: Engine/hosting, formats and glossary management - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:subprocessors Subdienstleister & Verarbeitungsregionen | MeinGPT. The register separates Germany platform storage, France backups and provider-specific EU inference regions. - URL: https://meingpt.com/sicherheit/subdienstleister - Locator: Register by function and processing region - Evidence kind: vendor-claimed - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:config Workspace-Konfiguration | MeinGPT Docs. Admins can restrict models by teams or people and supply workspace-wide instructions; denied API models return an error. - URL: https://meingpt.com/docs/admin-guide/workspace-configuration - Locator: Model access, default settings and additional chat context - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:limits Nutzungslimits | MeinGPT Docs. Personal category budgets block new use at zero; running meeting bots can overrun limits and scheduled runs are outside personal categories. - URL: https://meingpt.com/docs/admin-guide/usage-limits - Locator: Usage levels, meeting limits and user consumption table - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:api Einführung | MeinGPT Docs. Chat, embeddings, assistants, workflows and transcription APIs use bearer keys and published token rate limits; translation is UI-only. - URL: https://meingpt.com/docs/api - Locator: API overview, rate limits and errors - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### t3-chat:t3-chat:faq-usage T3 Chat FAQ. Base refills every four hours and Overage monthly; usage reserves then settles request cost and blocks until refill when both are depleted. Example20/200 points are illustrative, not published quota sizes. - URL: https://t3.chat/faq - Locator: How does the usage meter work? - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### t3-chat:t3-chat:r2-premier-buckets Premier usage-limit support. The identifiable vendor reply specifically applies the FAQ bucket mechanics to a Premier account. - URL: https://feedback.t3.chat/p/quickly-reaching-usage-limits-on-premier-plan - Locator: Mark Florkowski Team reply explaining exhausted monthly overage and four-hour bucket - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:pools Datenquellen & RAG | MeinGPT Docs. Cloud pools index six source types in Germany, refresh every 15 minutes and share a 300 GB raw-data budget. - URL: https://meingpt.com/docs/integrations/data-pools-rag - Locator: Cloud-source table, connector comparison and normal-chat search - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:chat Eingabefeld | MeinGPT Docs. Chat has stop, model choice, file upload, dictation and sharing links, with 50 files and 50 MB per non-image file. - URL: https://meingpt.com/docs/platform/chats/eingabefeld - Locator: Input controls, file limits, dictation and sharing - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:flow Datenfluss bei MeinGPT Chats | MeinGPT Docs. Chat deletion archives content; prompts have provider-dependent processing, and platform data stays until instructed deletion. - URL: https://meingpt.com/docs/privacy-security/faq-chats-datenfluss - Locator: Processing steps, providers and deletion rights - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:sandbox Code-Sandbox | MeinGPT Docs. Python/JavaScript analysis and Office generation run in an isolated Unikraft Cloud VM in Frankfurt, including for on-premise customers. - URL: https://meingpt.com/docs/platform/code-sandbox - Locator: Execution flow, hosting, lifetime and limitations - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:assistants Assistenten | MeinGPT Docs. Assistants combine model, instructions, files, pools, connectors, tools and memory. - URL: https://meingpt.com/docs/platform/assistants - Locator: Creation guide and editor sections - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:projects Projekte | MeinGPT Docs. Projects share instructions and workflows while each member keeps a private chat history. - URL: https://meingpt.com/docs/platform/projects - Locator: Collaboration, roles and deletion - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:m365 Microsoft 365 | MeinGPT Docs. Separate Outlook, Teams, SharePoint and OneDrive connectors run under user OAuth; Outlook can write, the other three read. - URL: https://meingpt.com/docs/integrations/connectors/microsoft365 - Locator: Overview, setup, tenant consent and permission model - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:pool-controls Sucheinstellungen am Datenpool | MeinGPT Docs. Source-permission import is off by default and request-enabled for SharePoint and Drive only; answers always show sources. - URL: https://meingpt.com/docs/integrations/data-pool-search-settings - Locator: Settings table and source-permission section - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:memory Memory | MeinGPT Docs. Memory persists per user across enabled assistants; users can inspect/delete it and admins can disable the tool. - URL: https://meingpt.com/docs/platform/memory - Locator: How memory works, management and admin controls - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:preferences Präferenzen | MeinGPT Docs. Users can set persistent system instructions and a platform-language preference. - URL: https://meingpt.com/docs/platform/preferences - Locator: Chat settings and appearance - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:roles Rollen & Zugriffsmodell | MeinGPT Docs. Admin, Member and Viewer roles differ in settings access and seat use; admins see all team-shared assets. - URL: https://meingpt.com/docs/admin-guide/roles-and-access-model - Locator: Role table, admin exception and support visibility - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:permissions Berechtigungen | MeinGPT Docs. Support-enabled permissions control creation, sharing and app use per person or team. - URL: https://meingpt.com/docs/admin-guide/permissions - Locator: Availability, permissions table and app restrictions - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:prompts Prompt-Bibliothek | MeinGPT Docs. Saved prompt folders have private, workspace or department visibility. - URL: https://meingpt.com/docs/platform/prompt-library - Locator: Core functions, visibility and sharing - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:web Websuche | MeinGPT Docs. Linkup powers chat search and a multi-step Deep Research app with cited sources. - URL: https://meingpt.com/docs/platform/web-search - Locator: Three routes, query creation, Linkup and research app - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### t3-chat:t3-chat:faq-url T3 Chat FAQ. A new-chat URL selects a model, reasoning effort, search on/off, one to five searches, a profile and temporary/persisted storage mode. - URL: https://t3.chat/faq - Locator: Custom search URLs and query parameter table - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:workflows Workflows erstellen und ausführen | MeinGPT Docs. Project workflows support multiple models, variables, documents and downloadable Office, PDF or JSON output. - URL: https://meingpt.com/docs/platform/workflows-create - Locator: Workflow editor and output format - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:scheduled Geplante Aufgaben | MeinGPT Docs. Scheduled tasks are support-enabled beta with schedule, webhook and email triggers under the creator account. - URL: https://meingpt.com/docs/platform/scheduled-tasks - Locator: Beta availability, triggers and execution account - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:tools Tool-Zugriffsrichtlinien | MeinGPT Docs. Tool methods are intersected across admin and assistant policies; there is no interactive per-action permission dialog. - URL: https://meingpt.com/docs/admin-guide/tool-access-policy - Locator: Security warning and four enforcement layers - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:audit Compliance Controls | MeinGPT Docs. Administrative settings logs exclude chat contents and are retained indefinitely; Governance Outpost search records are retained 90 days. - URL: https://meingpt.com/docs/admin-guide/compliance-controls - Locator: Core controls and Outpost search log - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:outlook Outlook Add-in | MeinGPT Docs. The Outlook add-in requires Enterprise API enablement and a personal API key and works in the received-mail reading view. - URL: https://meingpt.com/docs/integrations/connectors/outlook-addin - Locator: Prerequisites, supported clients and central deployment - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### t3-chat:t3-chat:api-request Subscription-backed API request. Staff rejects the economics of applying a T3 Chat subscription to external coding-agent/API usage; an official general chat API is not established. - URL: https://feedback.t3.chat/p/allow-t3-chat-subscription-to-power-external-coding-agents-or - Locator: Closed status and Mark Florkowski Team reply - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:terms Allgemeine Geschäftsbedingungen | MeinGPT | KI-Use-Case-Plattform für den Mittelstand. German commercial terms distinguish SaaS from contract-agreed on-premise deployment and describe subscriptions, export and termination. - URL: https://meingpt.com/agb - Locator: Sections 1a, 2, 3, 5 and final provisions - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### t3-chat:t3-chat:mobile Mobile App. The vendor released an iOS TestFlight beta on September10,2026 and says Android beta is coming soon. - URL: https://feedback.t3.chat/en/p/mobile-app-77 - Locator: Mark Florkowski, Team, pinned September10,2026 announcement - Evidence kind: documented - Source date: 2026-09-10 - Retrieved: 2026-10-21 ### meingpt:meingpt:images Bilder Generieren | MeinGPT Docs. The image app supports FLUX, GPT Image and Gemini image models with reference images and downloadable files. - URL: https://meingpt.com/docs/platform/image-generator - Locator: Model list, references and output management - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:image-admin Bildmodelle verwalten | MeinGPT Docs. Admins enable image models and set a default; the table distinguishes manufacturer, host and EU/global processing. - URL: https://meingpt.com/docs/admin-guide/image-models - Locator: Image-model controls and complete table - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:meeting Meetings | MeinGPT Docs. Recordings and transcripts become searchable meeting notes; Hetzner, Gladia and Vertex EU form the processing path. - URL: https://meingpt.com/docs/platform/meetings - Locator: Formats/limits, data flow and prices - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:prices prices. The public calculator uses marginal seat tiers plus model consumption and optional Governance and Outpost fees. - URL: https://meingpt.com/preise - Locator: Annual price tiers, included functions, add-ons and model-cost sections - Evidence kind: vendor-claimed - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:sso Access & Login Security | MeinGPT Docs. OIDC, Google and Microsoft login, verified domains and additive Entra group synchronization are documented; SAML is explicitly excluded. - URL: https://meingpt.com/docs/admin-guide/access-and-login-security - Locator: Own identity provider, authentication methods and Entra synchronization - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### t3-chat:t3-chat:security T3 Chat security policy. The vendor asserts TLS, MFA support, secure sessions/access control and regular audits; no independent report or enterprise enforcement configuration is supplied. - URL: https://t3.chat/security-policy - Locator: Data protection, authentication and disclosure program - Evidence kind: vendor-claimed - Source date: 2025-02-14 - Retrieved: 2026-10-21 ### meingpt:meingpt:scim Entra ID SCIM-Provisionierung | MeinGPT Docs. SCIM v1 provisions users only; deactivated accounts retain data and release their seat. - URL: https://meingpt.com/docs/admin-guide/entra-scim-provisioning - Locator: SCIM versus SSO, prerequisites, effects and seat assignment - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:analytics KI-Cockpit | MeinGPT Docs. Adoption and cost reports include user, model, API/platform and team breakdowns with CSV and ID-based anonymization. - URL: https://meingpt.com/docs/admin-guide/ai-cockpit - Locator: Typical metrics, user/team costs and knowledge search - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:audit-export Änderungsprotokoll exportieren | MeinGPT Docs. Admins export filtered administrative changes as CSV or JSON, capped at 10000 entries, 20 MiB and two minutes. - URL: https://meingpt.com/docs/admin-guide/admin-settings-map - Locator: Export procedure, fields and limits - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### t3-chat:t3-chat:privacy T3 Chat privacy policy. The policy covers user uploads, named hosting/model providers, no AI training, US/international processing and general disclosure practices. - URL: https://t3.chat/privacy-policy - Locator: Sections1-7 and11 - Evidence kind: documented - Source date: 2025-05-31 - Retrieved: 2026-10-21 ### t3-chat:t3-chat:terms T3 Chat Terms of Service. T3 Tools Inc permits organizational acceptance but each account is personal; proprietary restrictions, cancellation/export responsibilities and California dispute terms apply. - URL: https://t3.chat/terms-of-service - Locator: Sections1-6,9,19 and21 - Evidence kind: documented - Source date: 2025-11-15 - Retrieved: 2026-10-21 ### meingpt:meingpt:dpa-page Auftragsverarbeitungsvertrag | MeinGPT | KI-Use-Case-Plattform für den Mittelstand. DPA 2.0.1 dated 24 September 2026 needs signature or recorded electronic acceptance; existing customers retain their agreed version. - URL: https://meingpt.com/avv - Locator: Version and effectiveness - Evidence kind: documented - Source date: 2026-09-24 - Retrieved: 2026-10-21 ### meingpt:meingpt:dpa MeinGPT DPA 2.0.1. The signed-form DPA prohibits training, lists conditional subprocessors, defines authority-request handling and requires return or deletion within 30 days. - URL: https://meingpt.com/legal/avv/v2.0.1.pdf - Locator: Version 2.0.1; sections 4, 5(5), 7, 9, 12, 13 and Annexes 3/4, pages 3-12 - Evidence kind: documented - Source date: 2026-09-24 - Retrieved: 2026-10-21 ### meingpt:meingpt:security Sind meine Daten sicher? | MeinGPT Docs. Security documentation states TLS 1.3, AES-256, customer-managed keys on request, and no bulk chat/file export. - URL: https://meingpt.com/docs/privacy-security/faq-data-security - Locator: Technical security, deletion and data portability - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:trust Sicherheitsnachweise & Trust Center | MeinGPT. ISO 27001 and a 2025 SySS penetration-test certificate are listed behind email-gated access. - URL: https://meingpt.com/sicherheit/nachweise - Locator: Certificates, assessments and document access - Evidence kind: vendor-claimed - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:prices-rev2 Preise für Plattform, Add-ons und KI-Nutzung | MeinGPT. The contracted platform is offered with optional Governance and Outpost modules; local files are Early Access and local MCP/PostgreSQL Private Beta. The price calculator requires a binding individual offer. - URL: https://meingpt.com/preise - Locator: Feature table: Outpost release states; Governance separate agreement; calculator explicitly nonbinding - Evidence kind: vendor-claimed - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:secrecy-contract-rev2 meinGPT terms: professional secrecy. Section12(4) itself imposes indefinite professional secrecy and text-form staff/subcontractor obligations when covered data are processed; it does not state a Governance prerequisite. A separate addendum is outside the verified base-terms finding. - URL: https://meingpt.com/agb - Locator: Section12(4), conditional duties for professional-secret data; no Governance prerequisite stated - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:billing-rev2 Abrechnung verwalten | MeinGPT Docs. Self-service Business is EUR29 per user/month excluding VAT, billed monthly, with unlimited standard-model use; premium/reasoning, API and meeting-bot use are separately metered. - URL: https://meingpt.com/docs/admin-guide/billing - Locator: Workspace types; Business upgrade card; standard-model inclusion - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### t3-chat:t3-chat:overage Overage bucket refill option. The vendor says top-up refill has not been offered since the new billing model. - URL: https://feedback.t3.chat/p/overage-bucket-refill-option - Locator: Mark Florkowski Team reply - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### t3-chat:t3-chat:premier Pro and Premier pricing discussion. The vendor confirms Premier USD50/month plus tax, immediate upgrades and end-period downgrades, and customer OpenRouter keys for direct usage billing. - URL: https://feedback.t3.chat/p/new-tier-in-between-pro-and-premier - Locator: Mark Florkowski Team replies on price, OpenRouter, upgrades and downgrades - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### t3-chat:t3-chat:price-visibility Paid price visibility. Staff says billing information is needed for final tax-inclusive pricing; no checkout was completed. - URL: https://feedback.t3.chat/p/ability-to-see-paid-plan-price-without-going-into-checkout - Locator: Mark Florkowski Team replies - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### t3-chat:t3-chat:byok-openai OpenAI BYOK. Staff confirms GPT5 BYOK and recommends fallback priority plus a verified OpenAI organization or removing the key; this is historical model-specific setup, not a complete current model list. - URL: https://feedback.t3.chat/p/enable-byo-api-key-for-gpt-5-and-gpt-5 - Locator: Mark Florkowski Team reply - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:ecb-fx ECB historical euro reference rates, 20 October 2026. ECB2026-10-20 reference rates:EUR1=USD1.1215=CHF0.9237, implying USD1=EUR0.89166206. - URL: https://www.ecb.europa.eu/stats/eurofxref/eurofxref-hist.xml - Locator: Cube time=2026-10-20, USD and CHF currency entries - Evidence kind: documented - Source date: 2026-10-20 - Retrieved: 2026-10-21 ### t3-chat:t3-chat:byok-router OpenRouter model coverage. The request is marked Completed but the staff reply says those routes do not use OpenRouter and universal rerouting requires future changes. - URL: https://feedback.t3.chat/p/let-openrouter-keys-work-with-more-models - Locator: Completed label versus Mark Florkowski Team reply - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### t3-chat:t3-chat:fx ECB reference exchange rates. EUR1 equals USD1.1215 and CHF0.9237. - URL: https://www.ecb.europa.eu/stats/eurofxref/eurofxref-hist.xml - Locator: Cube time2026-10-20 - Evidence kind: documented - Source date: 2026-10-20 - Retrieved: 2026-10-21 ### t3-chat:t3-chat:openai-price OpenAI API pricing. The shared direct-provider proxy uses USD0.10 input/USD0.50 output per million tokens; T3 availability of this exact route is unverified. - URL: https://developers.openai.com/api/docs/pricing - Locator: Standard short-context gpt-6-luna row - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### t3-chat:t3-chat:openai-model GPT-6 Luna documentation. The shared direct-provider proxy uses lowest effort none; this is not a verified T3 default or model-menu entry. - URL: https://developers.openai.com/api/docs/models/gpt-6-luna - Locator: Reasoning effort options - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:billing Abrechnung verwalten | MeinGPT Docs. A separate Business offer includes standard models; the trial lasts 10 days with EUR 5 credit and no payment details. - URL: https://meingpt.com/docs/admin-guide/billing - Locator: Workspace types, trial differences and Business upgrade - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:terms-20261026 Allgemeine Geschäftsbedingungen | MeinGPT | KI-Use-Case-Plattform für den Mittelstand. German commercial terms distinguish SaaS from contract-agreed on-premise deployment and describe subscriptions, export and termination. - URL: https://meingpt.com/agb - Locator: Sections 1a, 2, 3, 5 and final provisions - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-26 ### meingpt:meingpt:status status. A public service-status page exposes incident history. - URL: https://status.meingpt.com/ - Locator: Platform component and incidents - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### t3-chat:t3-chat:faq-support T3 Chat FAQ. The FAQ links its official feedback board and lists support email. - URL: https://t3.chat/faq - Locator: Feature requests and support - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21 ### meingpt:meingpt:academy Lernen & Entdecken | MeinGPT Docs. The learning area includes guided learning paths and business use cases. - URL: https://meingpt.com/docs/platform/academy - Locator: Learning area content and personalization - Evidence kind: documented - Source date: Not stated - Retrieved: 2026-10-21