API error reference
Find an error by area, check whether a retry helps, and follow the fix. Statuses were checked against the live API where a request could reproduce them.
Authentication
- Missing Authorization headerHTTP 401, `Token not provided`
- Invalid API keyHTTP 401, `Invalid token`
- Disabled or expired API keyHTTP 403 or token error
Client configuration
- Invalid API base URLConnection error or HTTP 404
- Duplicated /v1 pathUsually tolerated by TokenReply
Protocol
- Wrong API endpoint for the requestHTTP 404, `Invalid URL`
Request format
- Malformed JSON request bodyHTTP 400
- Missing model fieldHTTP 400
- Content-Type mismatchHTTP 400
Model routing
- Model not foundError code `model_not_found`
Access policy
- Model access deniedHTTP 403 or access error
Billing
- Insufficient balanceBilling or quota error
Key limits
- API key quota exceededQuota error
Rate limiting
- HTTP 429 rate limitHTTP 429
Upstream health
- Upstream provider unavailableUsually HTTP 5xx
Network and latency
- API request timeoutClient timeout or HTTP 504
Request size
- Context length exceededHTTP 400 or upstream error
Generation settings
- max_tokens is too smallTruncated or empty output
Streaming
- Streaming parser shows empty outputHTTP 200 with unusable client output
Request compatibility
- Unsupported API parameterHTTP 400