Errors and Limits

AnyInt publishes both bearer-style and provider-style auth patterns. If a request fails unexpectedly, verify that you used the right header family for the route you called.

Common HTTP outcomes

Code	Meaning	Typical action
`400`	Invalid request body or missing fields	Validate payload shape before retrying
`401`	Authentication failed	Check your API key and header style
`403`	Access blocked by subscription or policy	Check entitlement, model access, or org policy
`429`	Limit reached	Back off, retry later, or upgrade limits
`503`	Upstream temporarily unavailable	Retry with backoff or use a fallback route

Practical guidance

Header mismatches

AnyInt publishes both bearer-style and provider-style auth patterns. If a request fails unexpectedly, verify that you used the right header family for the route you called.

Model access issues

Before treating a response as a platform outage, confirm:

the model ID exists
the model is enabled for your account
your subscription or key policy allows it

Rate and usage limits

Limits can come from:

subscription caps
per-key restrictions
team-level policies
upstream provider throttling

Async workflows

For video and AI Music routes, do not treat the initial task response as completion. Poll the task route or wait for a webhook before assuming the asset is ready.

Error response examples

Authentication failure

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

Check that you are using the auth header documented for the endpoint family. For OpenAI-compatible and most media routes, use Authorization: Bearer <ANYINT_API_KEY>. For Anthropic-compatible routes, use x-api-key: <ANYINT_API_KEY> with the documented compatibility headers.

Rate or quota limit

{
  "error": {
    "message": "Rate limit exceeded",
    "type": "rate_limit_error",
    "code": "rate_limit_exceeded"
  }
}

Back off before retrying. For user-facing products, show a temporary capacity or quota message instead of retrying in a tight loop.

Invalid request

{
  "error": {
    "message": "Missing required field: model",
    "type": "invalid_request_error",
    "param": "model"
  }
}

Fix invalid request parameters before retrying. Repeating the same invalid request will not succeed.

Retry guidance

Situation	Retry?	Guidance
Invalid JSON or missing required field	No	Fix the request body first.
Invalid API key	No	Rotate or correct the key before retrying.
Model not available to the account	No immediate retry	Call Models API and choose an available model.
Rate limit or quota exhausted	Later	Use exponential backoff or reduce concurrency.
Temporary upstream failure	Yes	Retry with exponential backoff and a maximum retry count.
Interrupted stream	Usually	Retry from the application level if the operation is safe to repeat.
Async task still pending	Poll, do not recreate	Continue querying the task route instead of creating duplicate tasks.

Family-specific notes

Anthropic-compatible routes may return request-shape errors if anthropic-version is missing
OpenAI-compatible chat returns stream chunks when stream is true
AI Music has additional task and callback status codes documented in its own section

On this page