Skip to main content

OpenRouter onboarding

Serenity Edge is being onboarded as a provider on OpenRouter. It is not live yet: today you cannot select Serenity Edge on OpenRouter. This page describes the status, what will work once it is live and what to use meanwhile.

Preview

OpenRouter has not enabled routing to Serenity Edge yet. When it does, the change will be announced in the release notes and on this page.

Status​

  • Serenity Edge side: ready. The model catalogue at GET /v1/models is published in OpenRouter's provider schema, and every chat model in it is marked ready.
  • OpenRouter side: not enabled yet. Until OpenRouter enables routing, requests sent through OpenRouter are not served by Serenity Edge.

What will work once it is live​

Provider and models​

Serenity Edge will appear on OpenRouter as the provider Serenity Edge. The catalogue offered is the chat models on Models, the same GET /v1/models document that OpenRouter reads. The exact model slugs on OpenRouter, and which of the Orion ids are exposed there, are being finalised with OpenRouter.

Decision models and the models of embeddings, rerank and redaction are not in that catalogue. They need a Serenity Edge API key.

Pinning the provider​

Once Serenity Edge is live on OpenRouter, pin the provider in your request so that OpenRouter does not fall back to another one:

{
  "model": "qwen/qwen3.8-27b",
  "provider": {
    "order": ["Serenity Edge"],
    "allow_fallbacks": false
  },
  "messages": [{"role": "user", "content": "Hello"}]
}

You will send the request to https://openrouter.ai/api/v1/chat/completions with your OpenRouter key. With allow_fallbacks: false, a request that Serenity Edge cannot serve at that moment fails instead of being routed elsewhere. Keep it false if data residency is why you chose Serenity Edge.

What will be different from direct access​

  • Pricing. The catalogue that OpenRouter reads carries Serenity Edge's published prices. OpenRouter bills you under its own terms, including its platform fee.
  • Rate limits. OpenRouter's account limits apply first; Serenity Edge's per-provider limits apply after.
  • Data. OpenRouter processes the request before forwarding it. From Serenity Edge onwards the data policy applies unchanged: prompts and completions are never written to disk or logs, prompt prefixes may stay in a volatile memory cache for at most 24 hours, and inference runs in Spain.
  • Features. Streaming, tool calling and structured outputs are declared per model in the catalogue, as listed on Models.

What to use meanwhile​

  • Serenity Star account. API keys and prepaid credits for direct access at https://api.serenityedge.ai/v1, including the Orion ids.
  • cortecs or EUrouter. If you prefer to reach Serenity Edge from a router or marketplace, these are available now, with that platform's own account, key and billing.
  • Enterprise. A contract with committed capacity and a service level agreement.
Last updated on