Is OpenRouter Down? Live status & outage reports

Service info and troubleshooting

API Key Not Working or Rejected

If requests keep returning 401 or 403, first check that you're passing the key correctly in the Authorization header as 'Bearer sk-or-...'. A common mistake is copying the key with a trailing space or missing the prefix entirely. Regenerate the key in your dashboard if it still fails — old keys sometimes get invalidated after account changes.

Model Returns Empty or Truncated Response

This usually means you've hit the context limit or the model cut off due to max_tokens being set too low. Check your request parameters: set max_tokens to a higher value or leave it unset to let the model decide. Some models on the platform also have hard caps — check the model card before sending long prompts.

Requests Timing Out

• Increase your client timeout to at least 60–120 seconds; large models can take that long on the first token
• If you're using streaming, make sure your client handles chunked responses properly
• Retry with a smaller or faster model to isolate whether it's a model-specific delay
• Check your local network — a flaky connection breaks streaming silently

Credits Not Updating After Payment

Payment goes through but the balance stays at zero. Wait two to three minutes and hard-refresh the billing page. If it still hasn't updated, check your email for a payment confirmation. Stripe sometimes delays the webhook. If the charge is confirmed but credits are missing, contact support with the transaction ID — they resolve it quickly.

Rate Limit Errors (429)

You're sending requests faster than your tier allows. Either add exponential backoff in your code or upgrade your account limit. OpenRouter enforces per-minute and per-day limits separately, so you can hit a daily cap even if your per-minute rate looks fine.

Interface Freezes on Model Selection

Hard-refresh the page with Ctrl+Shift+R to clear cached JS. If the playground keeps hanging when switching models, try a different browser — there are known issues with Safari and some Chromium builds when the model list is large.

Responses Not Streaming in Real Time

If you set stream: true but still get the full response at once, your HTTP client might be buffering. Disable response buffering explicitly, or check if a proxy sits between you and the API — corporate firewalls often buffer chunked transfer encoding.