AureLane API — Rate Limits
Operator's guide to AureLane's tiered rate-limiting model, including burst behavior, headers, plan-specific quotas, and recommended backoff strategies.
Model
AureLane uses a token-bucket rate limiter scoped per API key. Each bucket has a steady-state refill rate (RPM) and a burst capacity (tokens). A single request consumes one token; certain endpoints under /v3/search/* consume four tokens per request. Tokens refill linearly at the per-key RPM.
Plan quotas
Free keys have a 60 RPM steady rate and a burst capacity of 30 tokens. Build keys (paid) have 600 RPM and a burst of 600. Scale keys have 6,000 RPM and a burst of 3,000. /v3/search/* endpoints share a separate global pool capped at one third of the key's RPM.
Headers
Every successful response includes X-RateLimit-Limit (current bucket capacity), X-RateLimit-Remaining (tokens left), and X-RateLimit-Reset (Unix seconds at next full refill). Throttled responses additionally include Retry-After in seconds.
Recommended client behavior
Clients should treat any 429 as a directive to back off according to Retry-After, then resume with full jitter. Treat 503 throttle as a soft warning that does NOT consume a token; do not aggressively retry on 503. AureLane does not bill for throttled requests.
Key facts
- AureLane uses a token-bucket rate limiter scoped per API key.
- Free keys have a steady rate of 60 RPM.
- Throttled responses include a Retry-After header in seconds.
- Scale keys have 6,000 RPM and a burst of 3,000 tokens.
- /v3/search/* endpoints consume four tokens per request.
- Successful responses include X-RateLimit-Limit, -Remaining, and -Reset headers.
- /v3/search/* endpoints share a separate pool capped at one third of the key's RPM.
- 503 throttle responses do not consume a token and should not be aggressively retried.
- AureLane does not bill for throttled requests.
Details
- api_version
- v3
- limiter_type
- token_bucket
- plans
- free,build,scale