fixerror.dev

Anthropic Errors

Anthropic Claude API errors: invalid_request_error (400), authentication_error (401), permission_error (403), rate_limit_error (429), api_error (500), overloaded_error (529).

Anthropic uses typed error objects: `{type, message}`. Rate limits are tier-based RPM/TPM/ITPM (input tokens per minute). `overloaded_error` (529) is server-side capacity — retry with backoff. Long messages (>200k tokens) require Claude with extended context. Tool use errors require strict JSON Schema in the tool definition.

Authentication (1)

authentication_error 401

Invalid API Key

The `x-api-key` header you sent doesn't match an active Anthropic API key — usually because the env var isn't loaded, the key was rotated or revoked, you're using a Workspace key in the wrong workspace, or a wrong-provider key (Bedrock or Vertex) was sent to the direct Anthropic API.

Server errors (1)

overloaded_error 529

API Overloaded

Anthropic's infrastructure is at capacity for the model you requested. This is server-side, not a problem with your code or your account — Claude is experiencing a traffic spike or capacity event and rejecting requests until load eases.

Rate limits (1)

rate_limit_error 429

Too Many Requests

You exceeded one of Anthropic's per-minute caps for the model and tier — RPM (requests/min), ITPM (input tokens/min), or OTPM (output tokens/min). Anthropic enforces all three independently and you can hit any one without breaching the others.

Official docs

https://docs.anthropic.com/en/api/errors

Status page

https://status.anthropic.com/

Support

https://support.anthropic.com/

Related services

OpenAI

GPT API