# Rate limits

| Limit | Default | Notes |
| --- | --- | --- |
| Token route | 10 / minute per client | Cache tokens. |
| Per API client | 600 / minute, bursts up to 100 / second | Raised on request for backfills. |
| Per company | 1,200 / minute across all clients |  |
| Replace pricing rules | Counts as 10 requests | Bulk write. |

Every response tells you where you stand:

**Headers**

```
RateLimit-Policy: "client";q=600;w=60
RateLimit: "client";r=412;t=23
```

`r` is the number of requests left and `t` the seconds until the window resets. Slow down before `r` reaches 0.

Over the limit you get `429 rate_limited` with `Retry-After` (seconds) and `error.details.retryAfter`. Don't retry sooner: repeated early retries extend the block. A write refused with 429 never ran, so retry it with the same `Idempotency-Key`.

> Backfilling history? Use the bordereaux or the change feed with `size=100`, and ask your account manager to raise your limit for the backfill window.