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

# Limits

> Understand provider-driven rate limits and how to build resilient retry behavior.

The Gateway does not currently enforce a separate platform-level request cap. Most throttling comes from upstream providers.

## How limits work

* Requests can be rate-limited by the routed provider for the selected model.
* BYOK traffic uses the limits tied to your provider account.
* Gateway routing and fallback can reduce failures, but limits can still surface as `429`.

## Handling `429` responses

* Respect `Retry-After` when present.
* Use exponential backoff with jitter.
* Set a maximum retry count and fail gracefully.

```ts theme={null}
async function retryWithBackoff(run: () => Promise<Response>, maxRetries = 4) {
  for (let attempt = 0; attempt <= maxRetries; attempt += 1) {
    const response = await run();
    if (response.status !== 429) return response;

    const retryAfterHeader = response.headers.get("retry-after");
    const retryAfterMs = retryAfterHeader
      ? Number(retryAfterHeader) * 1000
      : Math.min(1000 * 2 ** attempt, 8000);

    await new Promise((resolve) => setTimeout(resolve, retryAfterMs));
  }

  throw new Error("Rate limit retries exhausted");
}
```

## Monitoring

* Track `429` rates by endpoint and model.
* Watch fallback frequency to identify provider pressure.
* Use dashboard metrics and your app logs together for incident triage.

## Related pages

* [Authentication](./authentication.mdx)
* [Parameters](./parameters.mdx)
* [Errors and Debugging](./errors.mdx)

If you are implementing retries as an agent:

* Use repository skills for safe retry loops and timeout control.
* Keep retry policy configuration centralized (attempt count, base delay, max delay).
* Never retry non-idempotent requests without explicit product approval.