OpenRouter is a unified API gateway from OpenRouter, Inc. that routes requests to hundreds of AI models from OpenAI, Anthropic, Google, Meta, DeepSeek, and others through a single OpenAI-compatible endpoint. Developers use it for one key with automatic fallbacks, and apps such as JanitorAI, SillyTavern, and Cline rely on it as their backend, so when OpenRouter fails, whole chains of tools break at once. If your requests are erroring, start with the live verdict at the top of this page.
Common reasons OpenRouter stops working
- A single upstream provider failing. Errors labeled as provider errors usually mean one upstream provider is saturated or down while the rest of the router is healthy.
- Rate limits on free model variants. Free variants are shared by huge numbers of users, so the upstream serving them gets temporarily rate-limited, returning 429 errors that normally clear within minutes.
- Free-tier daily caps. Accounts get 50 free-model requests per day, raised to 1,000 once you have bought at least 10 dollars in credits. Hitting the cap cuts chats off mid-session and is easy to mistake for an outage.
- Wider infrastructure incidents. OpenRouter runs behind Cloudflare, so a major Cloudflare or cloud platform incident can take every endpoint down at once, as a multi-hour outage in June 2025 showed.
- Credit and billing errors. When your balance runs out or goes negative mid-stream, requests fail with insufficient-credit errors that downstream apps often display as broken or empty responses.
- Login and dashboard failures. Sign-in runs on a separate authentication service, so the Chatroom and dashboard can be unreachable while existing API keys keep serving normally.
How to fix OpenRouter when it is not working
- Check the status at the top of this page first. If our check and reader reports show a broad outage, the problem is on OpenRouter’s side and waiting is usually the fastest fix.
- Switch models. Errors often come from one upstream provider, so trying a model served by a different provider quickly isolates the problem.
- Review your provider routing settings. A “No endpoints found” error means the model was delisted or your allowed and ignored provider preferences exclude every available route.
- Wait out free-variant limits. Rate-limit errors on free models usually pass within minutes, or move to the paid variant of the same model.
- Check your credit balance. Insufficient-credit failures surface as broken responses inside downstream apps. Top up, then retry the request.
- Test the API and the Chatroom separately. If the website or login fails, send a direct API request with your key; the two often fail independently.
- Rule out your own setup. Try another network or device, and check the app sitting in front of OpenRouter, since a wrong key or endpoint there produces identical symptoms.
For a fuller checklist that applies to any AI service, see our guide on what to do when an AI tool is down.
Is OpenRouter down, or is it just you?
Compare your experience with what others are reporting. If the verdict and reports at the top of this page show a spike, the outage is real and shared. If reports are quiet but your requests still fail, the cause is more likely local or partial: one upstream provider, your account limits, your credits, or the app calling OpenRouter on your behalf. The fastest single test is sending the same prompt to a model from a different provider; if that works, the problem is one route, not the gateway.
How we check OpenRouter
OpenRouter publishes an official status page at status.openrouter.ai, but no machine-readable feed that we can reliably read, so our verdict does not depend on it and you should check that page for provider-confirmed incidents. Instead, we run our own independent reachability check against OpenRouter from the outside, on a regular interval, and combine that signal with the crowd-sourced reports readers submit on this page. When many people report trouble at once and our check also fails, that is a strong sign of a genuine outage. You can read more about our approach on our methodology page.
Alternatives to OpenRouter during an outage
If OpenRouter is down and you need inference right away, Together AI and Groq both serve many of the same open models behind OpenAI-compatible APIs, while Hugging Face offers hosted inference across a broad model catalog. If the root cause is a wider infrastructure event, alternatives can be affected at the same time, so confirm your backup is healthy before switching. Keeping a second key configured in advance turns an outage into a one-line change instead of a scramble.
Why an OpenRouter failure is rarely the whole gateway
OpenRouter sits between your application and dozens of upstream providers, so most errors are narrower than they first appear. A failing request is often hitting one saturated provider route while every other route keeps serving, which is why many “OpenRouter is down” reports trace back to a single provider’s outage and vanish the moment you switch models. Free model variants add more confusion, because their strict request caps and shared upstream capacity produce rate-limit errors that are policy, not downtime.
The platform’s own components also fail independently. Login for the dashboard and Chatroom runs on a separate authentication service, so the website can lock you out while API keys keep working. At the other extreme, the whole service sits behind Cloudflare, so the rare true global outages tend to arrive with major Cloudflare incidents and take every endpoint down at once. OpenRouter’s own status page even tracks its homepage by region separately from the API.
- One model returning provider errors while others respond
- Rate-limit errors on free model variants at peak hours
- Chatroom or dashboard login failing while API keys work
- Every endpoint down at once during a Cloudflare incident