# TrustedRouter TrustedRouter is an OpenAI-compatible AI gateway for routing requests across hundreds of models and providers. It is designed for developers and teams that need configurable privacy routing, provider failover, and a verifiable prompt path. - Findings24: - ウェブサイト: - 料金: Freemium - カテゴリー: AIゲートウェイとルーティング, LLM API ## TrustedRouterとは？ TrustedRouter is an OpenAI-compatible AI gateway that routes applications to hundreds of open-weight and frontier models across more than 100 providers. Developers can preserve OpenAI-shaped calls and migrate by changing the API base URL; the site also documents an Anthropic-compatible endpoint and agent setup for Claude Code, Codex, Cursor, and similar tools. The service separates its production prompt path from the dashboard and billing control plane. Requests pass through an attested gateway running in a Trusted Execution Environment, with published evidence connecting the running image to public source. TrustedRouter states that prompt and output content are not logged and that the gateway fails closed when attestation cannot be verified. Routing aliases let teams choose a policy as well as a model. Available examples include `trustedrouter/auto` for provider rollover, `trustedrouter/zdr` for zero-retention routes, `trustedrouter/e2e` for confidential provider routes, and `trustedrouter/eu` for Europe-focused routing. BYOK is supported for customers that need to use existing provider commitments or rate limits. ## TrustedRouterでできること ### OpenAI-compatible gateway Use the endpoint at \`https://api.trustedrouter.com/v1\` with existing OpenAI SDK applications, chat or Responses calls, and many compatible agents. The documentation also lists an Anthropic-compatible base URL. ### Policy-based model aliases Select routing behavior through named aliases such as \`trustedrouter/auto\`, \`trustedrouter/zdr\`, \`trustedrouter/e2e\`, and \`trustedrouter/eu\`, or choose a direct model ID when exact selection is required. ### Attested, no-content-log prompt path The production gateway runs in a Trusted Execution Environment and publishes evidence connecting its running image to public source. TrustedRouter states that prompt and output bodies are excluded from control-plane and metadata systems. ### Provider failover and multi-cloud routing Health measurements and explicit fallback support routing across providers and GCP, AWS, and Azure. The service states that it does not silently fall back to an unverified prompt path. ### BYOK and published usage pricing Bring provider keys for existing commitments or enterprise rate limits. Prepaid text and embedding routes are billed at provider cost plus 5.5%, while asynchronous video generation is billed at the provider quote plus 20%; there is no monthly subscription. ### Open-source infrastructure and trust evidence The site identifies public frontend, backend, gateway, infrastructure, SDK, and deployment repositories, alongside public status, latency, availability, and attestation information. ## 利用シーン ### Migrate an existing OpenAI-based application Teams can create a key, change the base URL, run the documented PONG smoke test, and continue using OpenAI-shaped chat or Responses calls while selecting models from the catalog. ### Route sensitive or regulated-document workloads Legal and other sensitive-data applications can use the zero-retention or confidential routing aliases. The customer case study describes document extraction, retrieval preparation, and multi-stage legal reasoning for litigation materials. ### Run coding agents through one model gateway Developers using Claude Code, Codex, Cursor, eval runners, or other compatible agents can configure the documented base URL and environment variables, then choose a privacy, regional, or fallback policy. ### Keep existing provider commitments Organizations with provider credits, negotiated rates, or enterprise limits can use BYOK while retaining TrustedRouter's unified API, routing controls, and gateway trust boundary. ### Test model and provider alternatives Teams can compare direct model IDs and measured routes across a broad catalog, using \`trustedrouter/auto\` for healthy-provider rollover or cost-oriented route pools where appropriate. ## よくある質問 ### How do I migrate an existing application? Create a TrustedRouter API key, set the base URL to \`https://api.trustedrouter.com/v1\`, and run the documented PONG smoke test. Existing OpenAI SDK applications can then keep their current call structure while selecting a catalog model or routing alias. ### Does TrustedRouter store prompts or outputs? TrustedRouter states that prompt and output content is not logged and is excluded from its control-plane and metadata systems. Any content export is described as requiring explicit destination-level configuration. ### What happens if gateway attestation fails? The service states that the gateway fails closed rather than silently routing prompts through an unverified fallback path. ### Can I use my own provider API keys? Yes. BYOK is available for providers where a team already has committed spend or enterprise rate limits, while TrustedRouter supplies the unified API and routing layer. ### How is usage billed? There is no monthly subscription or seat fee. Prepaid text and embedding routes are charged at provider cost plus 5.5%, subject to a stated $0.01 per million token floor; asynchronous video generation is the direct provider quote plus 20%. BYOK and usage-based billing are also supported. ## クイック情報 - **Product type**: OpenAI-compatible AI gateway and model router - **Model and provider access**: 600+ models and 100+ providers, according to the site - **API endpoints**: OpenAI-compatible \`/v1\` endpoint plus a documented Anthropic-compatible endpoint - **Privacy model**: Attested gateway; no prompt or output content logs stated by the provider - **Billing**: No subscription or seat fee; prepaid, BYOK, and usage-based options - **Source domain**: trustedrouter.com