FAQ: Rate Limits

The server returns an error response immediately — requests are not queued. The AI client will surface this as an error message. You must wait for the sliding window to reset before retrying.

1. Per-minute limit resets after 60 seconds from the first request in that window.
2. Per-hour limit resets after 3600 seconds from the first request in that window.

Per InSync account. If you use the same InSync account from multiple devices or clients simultaneously, all requests count against the same limit. Limits are not split across devices.

Under normal usage, the per-hour limit supports approximately 300 questions per hour. Complex questions requiring multiple sequential tool lookups will consume more requests.

The AI client automatically uses parallel execution when possible — running multiple independent tools in a single request — and batches inputs within tools where supported. You do not need to request these optimizations explicitly.

Yes — every call to insync_assistant counts as a request. However, cached plans return faster, so you get results more quickly.

Enterprise customers can request increased rate limits. Contact InSyncAI Support at aisupport@insyncanalytics.com or through your account manager to discuss enterprise rate limit tiers.

No proactive warning is sent. The limit is enforced at the moment a request would exceed the threshold. If you are building an application or workflow that makes many calls, track your usage proactively.