Developers

Rate limits

The actual numbers, so you can size your integration before you write it rather than after.

Endpoint	Limit	Notes
Dashboard and authenticated API	240 requests per minute	Per workspace. Comfortably above interactive use; a backfill should pace itself below it.
Public chat endpoint	Per bot, configurable	Each assistant carries its own per-minute limit so one busy site cannot starve another. An adaptive bot challenge sits in front of it.
Sign in	10 per minute	—
Registration	20 per hour	—
Password reset	5 per hour	—
One-time code request	5 per hour	Verification of the code itself allows 20 per hour.
Team invitations	30 per hour	—
Test messages from the dashboard	20 per minute	—

When you hit one

Two different errors — tell them apart

Wait the seconds in the Retry-After header, then back off with jitter so clients don't re-synchronise.

A billing boundary, not a speed limit — retrying won't help until the cycle resets or you top up.

Retrying a 402 with backoff is the most common integration bug here — it waits for something that was never going to change on its own.

Bulk work

The API isn't billed per request, so there's nothing to gain by going faster.

A retry after a timeout returns the original result, not a second record.

The numbers are defaults, not a policy — a legitimate workload is worth an email.