Skip to main content

Confirm

Are you sure?

Changelog · September 20, 2026

Parallel requests per plan

A crawl of 5,000 pages costs the same credits however fast it runs; the higher plans finish it sooner.

By · Last updated: September 2026

TL;DR

Plans now sell speed as well as volume: Free runs 1 request at a time, Starter 3, Pro 8 and Scale 20. The limit rides on every response as X-Concurrency-Limit, a /scrape counts as one request, and a cache hit takes no slot at all.

Free plan, no credit card. 1,000 credits a month.

Credits say how much a month buys; they say nothing about how fast you get through it. X-Concurrency-Limit on every response tells a worker pool how wide to be, so it can size itself from the plan instead of discovering the limit by being turned away.

Over the limit, a request is refused with concurrency_limit — distinct from quota_exceeded and rate_limited, because the fix is different: it clears when one of your own requests finishes. There is no Retry-After, because there is no honest number to put in it. Nothing is spent on a refusal.

Waiting on a slow site that we are pacing does not hold one of your slots.

FAQ

Frequently asked questions

Does a /scrape use several parallel slots?
No. A scrape is one request however many operations it runs, because they come from one page visit.
Do cache hits count toward the parallel limit?
No. A cache hit does no work, so it takes no slot.
How do I know my plan's limit from code?
Read the X-Concurrency-Limit header, which is on every response.

Try it on the free plan.

Free plan, no card. Confirm your email and your API key is live — you'll be making real requests in minutes.