OpenRouter-compatible. Attested. 99.9% router availability.

OpenRouter-Compatible, But Verifiable

Change base_url, keep your models, get a verifiable non-logging prompt path.

Verify trust

1 linebase_url migration

99.9%router availability

0prompt/output logs

OpenRouter-compatible, but verifiable.Same OpenAI-shaped client, different trust model.

One APIAll the LLMs, provably private.

99.9%Router availability above individual endpoints.

Open sourceBackend, config, bring-up, and UI.

Why switch

All the routing convenience. A prompt path you can verify.

TrustedRouter lets you route to any LLM very easily. Switch between all your LLMs, have higher uptime, have backups, and try out much cheaper models, saving you money without needing to trust TrustedRouter servers as an intermediary any more than you need to.

Many engineers are using OpenRouter but only for low-security non-sensitive data because they know it's not so safe to put sensitive data in this third-party router.

Drop-in routeOpenAI-compatible

client = OpenAI(
    base_url="https://api.quillrouter.com/v1",
    api_key="sk-tr-v1-..."
)

model = "trustedrouter/auto"

Migrate

Change one URL.

It is as simple as changing one URL, so one line of code.

Open migration guide

Verify

Check the exact code path.

The trust page lists the commit hash of what code exactly is running.

Verify trust

Route

Use more prompts safely.

They have a lot more other prompts that they would love to use something like OpenRouter for.

OpenRouterTrustedRouter

API shapeOpenAI-compatible routingOpenAI-compatible routing

Trust modelOpaque third-party routerOpen source plus attestation

Prompt handlingYou have to trust the routerWe never log your prompt or the output

Best fitLow-sensitivity routingPrivacy-sensitive agent and app traffic

Founder words

OpenRouter convenience for traffic you could not send before.

The most common feedback I hear from engineers is that they don't use OpenRouter at all because of their security and privacy concerns, especially for sensitive prompts, or they use it only for low-sensitivity prompts.

Then you can see it working flawlessly. You can measure the uptime, the latency, the time to first token, and see that it's all really good and fast.

Most importantly it's all open source software. Every part of the backend infrastructure, configuration, bring up, and UI is entirely open source.