> Documentation index: [/llms.txt](/llms.txt). Every page as one file: [/llms-full.txt](/llms-full.txt).

# Rate limits

Rate limits [#rate-limits]

Routes that start model runs or handle sign-in and invites are limited per minute.

Limits [#limits]

| Route                                                                                                                                                                                                       | Default limit | Counted per     |
| ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------- | --------------- |
| `POST /api/v1/agents-v2/agents/{agent_id}/run-stream/`                                                                                                                                                      | 15 per minute | User or API key |
| `POST /api/v1/agents-v2/agents/run-stream/`                                                                                                                                                                 | 15 per minute | User or API key |
| `POST /api/v1/agents-v2/runs/`                                                                                                                                                                              | 15 per minute | User or API key |
| `POST /api/v1/agents-v2/agents/{agent_id}/sandbox/files/save-to-files/classify/`                                                                                                                            | 15 per minute | User or API key |
| `POST /api/v1/chat-links/public/{token}/message/` and `…/message/stream/`                                                                                                                                   | 15 per minute | IP address      |
| `POST /api/v1/auth/token`, `GET /api/v1/spaces/invites/preview`, `POST /api/v1/spaces/invites/set-password`, `GET /api/v1/organizations/invites/preview`, `POST /api/v1/organizations/invites/set-password` | 10 per minute | IP address      |

* The window is 60 seconds. The four agent-run routes share one counter per caller.
* Platform keys are never counted.
* Limits apply on deployments that enable them, including Alomana's shared cloud; a dedicated deployment may leave them off.

When you hit a limit [#when-you-hit-a-limit]

The response is `429` with headers that say when to retry:

```http
HTTP/1.1 429 Too Many Requests
Retry-After: 42
RateLimit: "agent-run";r=0;t=42
RateLimit-Policy: "agent-run";q=15;w=60
```

```json
{
  "detail": "Too many requests",
  "code": "REQUEST_RATE_LIMITED",
  "limit_name": "agent-run",
  "limit": 15,
  "window_secs": 60,
  "retry_after": 42
}
```

1. Read `Retry-After` (seconds).
2. Wait that long, then retry the same request.
3. To start many runs, spread them so no more than the limit start within any 60 seconds.

A plan limit is a different error: `402` with code `QUOTA_EXCEEDED`. See [Errors](./errors).