TrustedRouter — End-to-End Encrypted Router for every LLM

One API. Every LLM. Provable privacy.

One OpenAI-compatible key for hundreds of models across Anthropic, OpenAI, Google Vertex, Gemini, DeepSeek, Mistral, Cerebras, and more — prepaid or BYOK. Every prompt is end-to-end encrypted, with 99.9% router availability.

99.9% availabilitytrustedrouter/auto fails over across providers for uptime higher than individual LLM endpoints.

End-to-End Encrypted RouterTLS terminates inside the attested gateway; the SaaS control plane never sees prompts.

No subscriptionsPrepaid credits, BYOK, and usage-based billing. No monthly plan required.

Hosted OSSHosted service backed by open-source control plane and attested gateway code.

Verifiable APIThe trust page publishes source commit, image reference, and image digest.

One-line migration

Change base_url. Keep your models.

client = OpenAI(
    base_url="https://api.quillrouter.com/v1",
    api_key="sk-tr-v1-..."
)
resp = client.chat.completions.create(
    model="trustedrouter/auto",
    messages=messages,
)

trusted-router-py trusted-router-js

Trust Boundary

no prompt logs

Production inference belongs on api.quillrouter.com. The SaaS control plane stores metadata and billing records only. Every model in the catalog — Anthropic Claude, OpenAI GPT, Google Vertex, Gemini, DeepSeek, Mistral, Cerebras — runs through the same attested non-logging gateway.

Open trust page

Payments

credits + BYOK

Stripe Checkout, PayPal, and Stablecoin checkout are wired for prepaid credits. Auto-refill keeps prepaid routes flowing once you've saved a Stripe card.

Production

multi-region

The attested gateway runs in 2 GCP regions with 99.9% router availability. Pick one explicitly via the region header or let the router pick the closest active one for you.

Inference, attested, in 2 regions

Decentralized low-latency routing. 99.9% router availability, higher than individual LLM endpoints.

us-central1 Iowa primary
europe-west4 Netherlands