TrustedRouter
AIOpenAI-compatible gateway to 600+ models across 90+ providers, routed through an attested enclave that logs no prompt or output content.
TrustedRouter
OpenAI-compatible gateway to 600+ models across 90+ providers, routed through an attested enclave that logs no prompt or output content.
About TrustedRouter
TrustedRouter is an LLM routing gateway that exposes 600+ open-weight and frontier models from 90+ providers behind a single OpenAI-compatible API, so migrating takes only a base_url change in an existing OpenAI SDK client. Its distinguishing claim is verifiable privacy: every request crosses a cryptographically attested gateway running in a Trusted Execution Environment whose hosted image digest can be matched against a public source commit, and the prompt path keeps no content logs — the operators state that even their own engineers cannot read requests. Instead of picking a model, callers can pick a routing promise through named aliases such as trustedrouter/auto for healthy-provider rollover, /zdr for zero-retention providers, /e2e for confidential compute, /eu for Europe-focused routes, and /synth for combining several models into one answer. Billing is prepaid credits or usage-based with no subscription, charging the provider price plus a 5.5% markup, and BYOK is supported for providers where a team already holds committed spend or enterprise rate limits. The production prompt path runs separately from the dashboard and billing control plane, and the source is public so the trust claims reduce to evidence customers can inspect themselves.
Screenshots
Key Features
Use Cases
Frequently asked questions about TrustedRouter
What is TrustedRouter?
OpenAI-compatible gateway to 600+ models across 90+ providers, routed through an attested enclave that logs no prompt or output content.
How does TrustedRouter work?
TrustedRouter works by combining One OpenAI-Compatible Endpoint: Point an existing OpenAI SDK client at api.trustedrouter.com/v1 and reach 600+ models across 90+ providers and three clouds without rewriting call sites., Attested No-Log Prompt Path: Requests cross a gateway running in a Trusted Execution Environment that publishes no prompt or output logs, with the image digest matchable to a public source commit before production traffic moves., Routing Promise Aliases: Named models like trustedrouter/auto, /zdr, /e2e, /eu and /synth make failover, zero-retention, confidential compute, EU residency and multi-model answers explicit at request time., Provider Failover and Regional Routing: Automatic rollover to healthy providers keeps one upstream outage from becoming an application outage, with region-constrained routes available., Bring Your Own Key: BYOK preserves existing committed-spend discounts and enterprise rate limits while still gaining the attested prompt path and unified API. to help users with Regulated Data Processing: Running legal, medical or financial documents through LLMs where a no-retention, inspectable prompt path is a client requirement., OpenRouter Migration: Moving an existing multi-model application to a cheaper gateway with one base_url change and no client rewrite., Cost Optimization Across Models: Routing routine traffic to open models like Qwen, GLM, DeepSeek, Gemma, Kimi or MiniMax while keeping frontier models available for hard requests., EU Data Residency: Constraining inference to European providers for workloads governed by EU rules., Production Reliability Engineering: Using provider failover so a single vendor outage does not take down a customer-facing AI feature..
What are the main features of TrustedRouter?
Key features include One OpenAI-Compatible Endpoint: Point an existing OpenAI SDK client at api.trustedrouter.com/v1 and reach 600+ models across 90+ providers and three clouds without rewriting call sites., Attested No-Log Prompt Path: Requests cross a gateway running in a Trusted Execution Environment that publishes no prompt or output logs, with the image digest matchable to a public source commit before production traffic moves., Routing Promise Aliases: Named models like trustedrouter/auto, /zdr, /e2e, /eu and /synth make failover, zero-retention, confidential compute, EU residency and multi-model answers explicit at request time., Provider Failover and Regional Routing: Automatic rollover to healthy providers keeps one upstream outage from becoming an application outage, with region-constrained routes available., Bring Your Own Key: BYOK preserves existing committed-spend discounts and enterprise rate limits while still gaining the attested prompt path and unified API..
Who is TrustedRouter for?
TrustedRouter is useful for anyone interested in Regulated Data Processing: Running legal, medical or financial documents through LLMs where a no-retention, inspectable prompt path is a client requirement., OpenRouter Migration: Moving an existing multi-model application to a cheaper gateway with one base_url change and no client rewrite., Cost Optimization Across Models: Routing routine traffic to open models like Qwen, GLM, DeepSeek, Gemma, Kimi or MiniMax while keeping frontier models available for hard requests., EU Data Residency: Constraining inference to European providers for workloads governed by EU rules., Production Reliability Engineering: Using provider failover so a single vendor outage does not take down a customer-facing AI feature..
How much does TrustedRouter cost?
TrustedRouter is a paid product. Check the official site for current pricing.
How do I get started with TrustedRouter?
Visit https://trustedrouter.com/ to sign up and explore TrustedRouter.
Explore more AI Ai Services tools
Browse all Ai Services tools →Compare TrustedRouter: vs Cadenya · vs OpenObserve · vs Experiential Labs · vs Dial