Errors

Consistent envelopes and actionable codes.

lmchat aims to return OpenAI-style error envelopes where possible. Your client should handle authentication errors, validation errors, rate limits, and upstream provider transient failures.

Error envelope

{
  "error": {
    "message": "Invalid API key",
    "type": "authentication_error",
    "code": "invalid_api_key"
  }
}

Common HTTP statuses

400 — invalid_request_error

Malformed JSON, missing required fields, invalid parameters.

401 — authentication_error

Missing/invalid bearer token.

403 — permission_error

Key lacks access to a model/provider/policy boundary.

429 — rate_limit_error

Throttled by lmchat or upstream provider; retry with backoff.

5xx — api_error

Transient platform/provider issues; safe to retry with bounds.