# TrustedTokens > TrustedTokens is a sovereign AI inference platform by TNG Technology Consulting GmbH. It serves leading open-weights language models on TNG-operated NVIDIA GPUs in German datacenters, with EU data residency, no prompt storage, no CLOUD Act exposure, and an OpenAI-compatible API. TrustedTokens is aimed at European businesses that need frontier LLM capabilities without sending data to US hyperscalers. Inference runs in Munich/Aschheim on TNG hardware, with scale-out capacity in Munich. The API is OpenAI-compatible and supports chat completions, responses, embeddings, and Anthropic Messages. Plans combine a monthly fee with a usage credit, and model usage is billed per token in EUR. ## Main benefits - **Sovereign EU inference** — models run on TNG-operated NVIDIA GPUs in German datacenters; data is processed within EU jurisdiction and is not exposed to the CLOUD Act. - **No prompt retention** — prompts and outputs are not stored or used to train models; only operational metadata is logged. - **OpenAI-compatible API** — use existing OpenAI SDKs and tooling with one API, one contract, and one bill. - **Open-weights choice** — frontier open-weights models and TNG fine-tunes, with context windows up to 1M tokens and no vendor lock-in. - **Enterprise-ready** — DPA available, EU payment processing via Mollie, verified business onboarding, and transparent EUR pricing. - **Transparent operations** — public model catalog, pricing, and platform activity data. ## Main pages - [Home](https://trustedtokens.eu/): product overview, main benefits, and current model news. - [Models](https://trustedtokens.eu/models): hosted model catalog with capabilities, context limits, lifecycle state, licenses, and per-model pricing. - [Pricing](https://trustedtokens.eu/pricing): subscription plans and per-token model rates in EUR, excluding VAT. - [Documentation](https://trustedtokens.eu/docs): API base URL, authentication, supported endpoints, code examples, and agent configuration. - [FAQ](https://trustedtokens.eu/faq): data processing, sub-processors, DPA, billing, cancellation, and business-customer eligibility. ## Models and pricing data - [Model catalog and pricing JSON](https://trustedtokens.eu/api/public/models): public, no authentication. Each model includes its name, mode (chat or embedding), capabilities, attributes, lifecycle, license, context limits, 7-day usage, and providers. `providers[].pricing` contains EUR prices per 1 million tokens: `input`, `output`, and `cache_read` for chat models, plus `price_update` for announced changes. Embedding models price input tokens only. - [Inference API OpenAPI spec](https://trustedtokens.eu/openapi.json): public OpenAPI 3.1 description of the supported inference surface: `/v1/chat/completions`, `/v1/responses`, `/v1/embeddings`, `/v1/models`, and `/anthropic/v1/messages`. - [Models page](https://trustedtokens.eu/models): human-readable model catalog and pricing. - [Pricing page](https://trustedtokens.eu/pricing): human-readable subscription plans and model pricing. ## Platform statistics - [Platform statistics JSON](https://trustedtokens.eu/api/public/platform-stats): public, no authentication. Returns platform-wide totals and roughly 8 weeks of history: total requests and tokens, success rate, throughput, weekly per-model token buckets, daily request and success-rate series, and per-model rankings. Responses are cached for about 10 minutes. - [Model usage in the catalog](https://trustedtokens.eu/api/public/models): each model's `usage_tokens_7d`, the platform-wide token volume for the last 7 days. ## Complete page list - [Sitemap](https://trustedtokens.eu/sitemap-index.xml): every route on the site, including account-flow pages. The curated links above are the public-content entry points; account, usage, activation, and verification pages require authentication or are part of the signup flow.