- Docs
- Support
- FAQ & error codes
FAQ and error codes
Start with the error code if a request failed. If your question isn't here, email [email protected] with the request ID from the error message.
Error codes
| Status · code | What happened | What to do |
|---|---|---|
| 400 | The request body is invalid, or the model rejected a parameter. | Read message; remove parameters the model doesn’t support. |
| 401 | The key is wrong, disabled, expired or over its spending limit. | Copy the key again without spaces; check it on the API keys page. |
403insufficient_user_quota | Your balance can’t cover the request. | Top up in the console. |
| 429 | Too many requests in a short time. | Retry with exponential backoff. |
503model_not_found | The model ID isn’t available on Airoutly. | Copy the exact ID from the models page. IDs are case-sensitive. |
| 500 · 502 · 504 | The model provider failed or timed out. | Retry. For long outputs, use streaming. |
Account and keys
Does one key work with every tool?
Yes. The same key works with every model, every protocol and every tool. Create separate keys only if you want separate spending limits or usage records.
I didn’t get the verification email
Check spam, then wait two minutes before requesting a new code. Emails come from [email protected]. Still nothing? Write to [email protected].
What if my key leaks?
Delete it on the API keys page right away and create a new one. A deleted key stops working immediately.
Billing
How am I charged?
Per token, at each model’s price on the models page, from a prepaid USD balance. Cached input is billed at the lower cache-read price. Every request is itemised in Usage logs.
In what order are credits and balance used?
Credits from redeem codes are used first, the one expiring soonest before the others. Your balance pays for the rest.
Requests
Which Base URL do I use?
OpenAI SDKs and tools: https://api.airoutly.com/v1. Anthropic SDKs and Claude Code: https://api.airoutly.com. Gemini SDKs: https://api.airoutly.com/v1beta.
Long requests time out
Non-streaming requests longer than about two minutes can time out at our edge. Turn on streaming; streamed replies aren’t affected.