Introduction
An OpenAI-compatible API that picks the right model for every request, and can score every token it returns.
Telluvian is an OpenAI-compatible API with two products on one endpoint. Routing picks which model answers each request. Verification can score every token it returns for hallucination, when you turn it on. Use either, or both.
The API is OpenAI-compatible. If you can call OpenAI, you can call this. Change the base URL and the key.
https://api.telluvian.ai/v1Routing
Gallery is our router, and you call it like a model. Send
telluvian/gallery-1 and each request goes to the cheapest model that we
expect clears the quality bar you set. The response reports which model ran.
See Gallery.
Model Select is the engine behind it. It reads each prompt and predicts how
well every model in the model zoo will handle it. You can also call it directly with
/v1/modelSelect to get the decision and make the model call yourself. See
Model Select.
You set the quality bar and the allowed models. See Pricing for the fees.
Verification
Verification returns a score for every generated token, not one number for the
whole response, so a score points at the specific claim that looks wrong. The
scores are not calibrated probabilities. Verification is off by default; turn it
on per request with include_scores: true. See Reading the scores.
Set up with a coding agent
Copy this prompt into your coding agent and it will make the integration in your project, following our agent guide.
Set up Telluvian routing in this project. 1. Read https://telluvian.ai/docs/agents/routing.md before changing anything. 2. Find where this project calls an LLM (OpenAI SDK or compatible). 3. Point it at https://api.telluvian.ai/v1 with model "telluvian/gallery-1". 4. Read the key from TELLUVIAN_API_KEY. Add it to .env.example; never hard-code it. If I don't have a key yet, tell me to create one at https://telluvian.ai/dashboard/keys. 5. Log the response's `model` field so I can see which model answered. 6. Make one test request and show me the result. Ask before changing production defaults. 7. Ask whether I want to use my own OpenAI, Anthropic, Google or OpenRouter key (BYOK; see https://telluvian.ai/docs/byok.md). Those keys are added in the dashboard at https://telluvian.ai/dashboard/provider-keys, never in this project's code or .env. If I do, log the response's `usage.byok_credential` to show which key served each request.
Paste it into Claude Code, Cursor, Codex or any coding agent, from your project folder. Agent without web access? or download AGENTS.routing.md.