> ## Documentation Index
> Fetch the complete documentation index at: https://developers.huechat.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Rate limits

> Per minute, per credential, announced on every response.

| Credential or surface     | Limit                                         |
| ------------------------- | --------------------------------------------- |
| Personal access token     | 1,000 requests per minute                     |
| Scoped API key            | The key's own `rate_limit_rpm` (1 to 100,000) |
| AI routes under `/api/v2` | Burst of 10, then 1 request every 2 seconds   |

Short bursts above the per-minute rate are tolerated up to about one sixth of
the limit.

## Headers

Every response carries the current window:

```
X-RateLimit-Limit: 1000
X-RateLimit-Remaining: 995
X-RateLimit-Reset: 1712345738
```

`X-RateLimit-Reset` is a Unix timestamp.

## When you are limited

```
HTTP/1.1 429 Too Many Requests
Retry-After: 60

{ "error": "Rate limit exceeded", "limit": "1000 requests per minute" }
```

Wait for `Retry-After` seconds, then back off exponentially if it happens
again. Reads that can be batched should be: one bounded list call beats a
poll per record.

<Tip>
  Move event-driven work to [webhooks](/webhooks). A subscription costs no
  requests at all.
</Tip>