# Rate Limits and Quotas

The IntelCenter API applies rate, quota, credit, and abuse controls to protect availability and enforce account allowances. Exact limits depend on the API key and capability.

## Controlled Operations

Controls may apply independently to:

- Structured search and detail retrieval
- Media access
- Geo Manifest traversal
- Natural Language Search
- Domain GenAI
- MCP tool calls

## Limit Response

A request blocked by a rate or usage control returns `429 Too Many Requests` with a documented error body.

```json
{
  "error": "quota_exceeded",
  "message": "This API key exceeded a usage or rate limit and the request was blocked."
}
```

## Retry Guidance

- Honor `Retry-After` when present.
- Use exponential backoff with jitter.
- Bound retry attempts.
- Cache safe retrieval results where appropriate.
- Reduce polling and use date filters or cursors.
- Do not automatically retry Domain GenAI as an idempotent request; a successful replay can create another turn and consume credits.
- Conversation deletion is idempotent and can be retried safely after a transport failure.

Contact IntelCenter support when an integration needs different account limits.