Rate limits

Airscale rate limits are endpoint-specific. Each endpoint can enforce a different request limit and time window, so use the value documented on the endpoint page you are calling.

​

Limits by endpoint

All numeric limits below apply per workspace unless the row names another scope. A shared limit is one combined allowance for the named routes. Find people shares its 6 requests per second with Count people. Find companies shares its 6 requests per second with Company filter values and Business intent topics. These are separate limits for the two endpoint families. Post search, Profile comments, Comment likers, Profile posts, and Company posts each have their own 180-request allowance; they do not share it with each other or with Post likers and Post commenters. Company lookalikes returns 429 when its rate or concurrency limit is reached. Honor Retry-After when present. Repeating a search can cause another charge; see the endpoint guidance. The job-change monitor endpoints share one allowance of 120 requests per fixed minute, counted per API key across all ten endpoints, and return Retry-After: 60 with 429. The new hire monitor endpoints and the post engagement monitor endpoints each have their own allowance of 120 requests per fixed minute per API key, shared across the endpoints of that API, and return Retry-After: 60 with 429. For bulk email, the allowance counts input items, not HTTP requests. An endpoint without a documented numeric limit should not be treated as unlimited; use bounded concurrency and handle 429 responses.

​

Handle rate-limit responses

When a request exceeds an endpoint’s limit, the API returns 429 Too Many Requests. Treat the status code as the rate-limit signal; do not depend on an error field or response header unless that endpoint page documents it. Only retry when the endpoint page says the operation is safe to retry. Use bounded exponential backoff: cap both the maximum delay and the total number of attempts, then surface a persistent 429 to your application.

​

Check an endpoint’s limit

Review the request limit on the endpoint page before setting concurrency: