Rate limits

Every operation has a rate class. A class sets one or two limits, and each limit counts your organisation's requests to that operation, whichever credential sends them. The marketplace catalogue, which needs no credential, counts per client IP instead.

Class	Limits	Used by
read	100 per second and 360,000 per hour	reads
write	50 per second and 36,000 per hour	creates, updates and actions
telemetry	10 per second and 180,000 per hour	positions, tracks and fuel levels
command	20 per second and 360,000 per hour	device commands
bulk	100 per hour	fuel statement imports
report	1,000 per day	report runs, a schedule run now
credentials	10 per minute	API key rotation and revocation, personal access tokens, webhook secret rotation
public	100 per second and 360,000 per hour, per client IP	the marketplace catalogue

The figures are the gateway's configuration, the same that Developer Studio shows. There are no per-key or per-plan limits. A sandbox is its own organisation, so its traffic counts apart from yours.

Over a limit

The gateway answers 429 with {"error_msg": "rate limit exceeded"}. Each response carries X-RateLimit-Limit, X-RateLimit-Remaining and X-RateLimit-Reset: wait until the reset before retrying, and spread steady work below the per-second limit.

Don't poll

• For geofence entries, alerts, stop completions, fuel flags and maintenance, subscribe to webhooks instead of listing.
• For positions, GET /v1/positions/latest returns many devices in one call.