Free LLM API errors · OpenRouter · checked 2026-11-13
OpenRouter "Provider returned error": fix
HTTP 429 · Provider returned errorOpenRouter passed your request to the provider that serves the model, and that provider refused it. With free models it is usually capacity: the free endpoint is saturated.
Our own router has received this error from OpenRouter in production.
Why it happens
- The provider serving the :free variant is at capacity. In our own traffic it arrives as a 429 and clears within minutes.
- It is not your own limit: your free-models-per-day count can be far from the limit when it happens.
- Popular free models hit it most at peak hours.
How to fix it
- Retry after a short pause (seconds to a few minutes), with backoff.
- Keep a list of free models and move to the next one on this error instead of retrying the same one forever.
- If the model also exists without :free, the paid variant is usually served with more capacity.
- Do not mark the model as broken for hours: it is a capacity problem and usually passes quickly.
Questions
- Does "Provider returned error" count against my free requests?
- We cannot see OpenRouter's counter from outside, so check it yourself: GET /api/v1/key shows how many free requests you have used today.
- Is "Provider returned error" the same as being rate limited?
- Not by you. Your own limit gives "Rate limit exceeded: free-models-per-day" or "…per-min". This message means the provider behind the model said no.
Sources
Read on 2026-11-13. Providers change limits and messages often; the linked pages are what applies.
Limits and fine print of OpenRouter's free tier: OpenRouter free API. Tired of handling these by hand? ZeroLimitAI is one OpenAI-compatible endpoint that moves to the next free model when one of these comes back, and can run on your own Groq or OpenRouter key.