Rate Limits

Requests are limited by authenticated API key. Read the effective policy instead of assuming a fixed contractual limit.

Authenticated responses include:

RateLimit-Policy: 120;w=60
RateLimit: limit=120, remaining=119, reset=60

GET /capabilities also reports the key’s active request-per-minute limit and content limits.

Rate-Limited Response

A rejected request returns 429 rate_limit_exceeded with Retry-After and Problem Details.

Recommended Client Behavior

• Honor Retry-After.
• Use exponential backoff with randomized jitter.
• Apply a total retry budget.
• Reuse the original idempotency key for a retried mutation.
• Avoid synchronized retries across workers.
• Queue high-volume submission instead of sending uncontrolled bursts.

Related Reference

• Get effective capabilities
• Idempotency and retries