Common Errors
If a client returns an error after connecting to yylx.io, start with the table below. It's organized by error code (error_code), error message (error_message), and explanation, covering the most common cases.
When an error occurs, first confirm the Base URL, API Key, and model name are correct, then use the table to locate the cause. Most errors come down to client configuration or current usage, and adjusting then retrying usually resolves them.
Error code reference
| Code | Error message (error_message) | Explanation |
|---|---|---|
400 / api_error |
API key group platform is not gemini |
The Key's platform doesn't match the requested endpoint. Make sure the Key type matches the endpoint — e.g. the gemini endpoint requires a gemini-platform Key. |
401 / api_error |
User account is not active |
The account is not activated or has been suspended. Check the account status, or contact the administrator. |
403 / api_error |
This group is restricted to Claude Code clients (/v1/messages only) |
This group only allows Claude Code clients. Switch to Claude Code, or change to a group/Key that supports your client in the console. |
403 / api_error |
Access denied. Your IP is <ip> |
Your current IP is blocked. If your own network looks fine and this persists, contact the administrator. |
404 / api_error |
Model "<name>" is not supported by any configured account in this group |
The model isn't supported by the current group. Check the full model name in the console — it's case-sensitive; don't use unpublished or misspelled names. |
429 / rate_limit_error |
Too many pending requests, please retry later |
Too many concurrent requests in a short window. Lower concurrency, or retry later. |
500 / api_error |
All available accounts exhausted |
Service capacity is temporarily tight and the request can't be handled right now — retry later. |
502 / api_error |
Upstream request failed |
The service is temporarily unavailable — retry later. |
502 / api_error |
Upstream stream ended without a response |
The response was interrupted before it started — send the request again. |
503 / api_error |
No available accounts: this group only allows Claude Code clients |
This group only admits Claude Code clients. Switch to Claude Code, or change to a group/Key that supports your client in the console. |
200 / api_error |
Recovered upstream error 502/403/429/500 (returned mid-stream) |
The response was interrupted while streaming and the content you received may be incomplete — send the request again. |
Troubleshooting tips
- Check the configuration first: confirm the Base URL, API Key, and model name all match the console.
- Read the error code: 4xx errors usually relate to the request itself or account/group configuration; 5xx errors are transient server-side states — retrying later usually works.
- Back off on retries: with 429 or 503, rapid retries make things worse — back off gradually before retrying.
- Keep the details when reporting: if retries don't help, include the request time, error code, and error_message so the issue can be located quickly.