free until your first top-up, then paid. GET /v1/limits returns yours. A call over a limit answers 429 and costs nothing.
Limits
A job start is
POST /v1/crawl, /v1/batch/scrape, /v1/extract or /v1/agent. A browser start is POST /v1/browser, a browser task or an interact call. The calls limit counts each endpoint on its own, so a busy scrape does not use up your search limit. The three other limits are shared by every call of their kind.
Without an API key you can call search, scrape, map and parse, over HTTP and MCP, up to 10 calls a minute and 50 calls a day per IP address, with no paging. Scrape with a parser works without an API key when the parser is marked keyless: true. Crawl, batches, extract, agent, browsers and monitors need an API key. See API keys. A first top-up lifts you to the paid limits.
Read your limits
GET /v1/limits returns your tier, limits and what is left this minute. It counts as one call. On the free tier, paidLimits shows the limits you get after a top-up.
Check calls running at once
GET /v1/concurrency-check is free. It returns how many of your calls are running and how many can run at once.
concurrency counts the check itself. Size your worker pool to maxConcurrency. GET /v1/team/queue-status shows your running and waiting jobs. See Usage and logs.
Check calls without an API key
GET /v1/keyless/eligibility needs no API key. It says whether your IP address can still make calls without an API key, and how many it has left.
Headers
Responses includeX-RateLimit-Limit, X-RateLimit-Remaining and X-RateLimit-Reset, the Unix time in seconds when the window resets. Without an API key they count the day’s calls. These headers came from a real scrape without an API key.
When you go over
A
rateLimited answer looks like this. This is the shape from the API reference, and the numbers are an example.