What is concurrency in a scraping API?
Concurrency in a scraping API is the number of requests or browser sessions an account may have running at the same time. It is not the same as a monthly allowance or a requests-per-second limit: concurrency counts unfinished work, while those other limits count volume or starts over a period. (ScraperAPI concurrency FAQ, Firecrawl pricing, and Firecrawl Browser Sandbox, checked 2026-07-26.)
Concurrency across the APIs we track
| API | Cheapest plan | Largest plan | Checked |
|---|---|---|---|
| AgentQL | 100 on Professional ($99.00 a month) | the same on every plan that publishes it | Oct 9, 2026 |
| Anchor Browser | 25 on Starter ($50.00 a month) | 200 on Growth ($2,000.00 a month) | Oct 9, 2026 |
| Browserbase | 25 on Developer ($20.00 a month) | 100 on Startup — Extract ($99.00 a month) | Oct 9, 2026 |
| BrowserCloud | 5 on Hobby ($19.00 a month) | 100 on Business ($249.00 a month) | Oct 9, 2026 |
| Browserless | 10 on Prototyping ($35.00 a month) | 100 on Scale ($500.00 a month) | Oct 9, 2026 |
| Desearch | 150 on Starter ($179.00 a month) | 600 on Scale ($1,599.00 a month) | Oct 9, 2026 |
| Firecrawl | 5 on Hobby ($19.00 a month) | 100 on Scale ($749.00 a month) | Oct 7, 2026 |
| HasData | 5 on Startup ($59.00 a month) | 500 on Growth 20M ($1,660.00 a month) | Oct 8, 2026 |
| Hyperbrowser | 25 on Startup ($30.00 a month) | 100 on Scale ($100.00 a month) | Oct 5, 2026 |
| Kernel | 10 on Hobbyist ($30.00 a month) | 150 on Start-Up ($200.00 a month) | Oct 9, 2026 |
| Lightpanda Cloud | 30 on Builder ($19.00 a month) | the same on every plan that publishes it | Oct 9, 2026 |
| Olostep | 150 on Starter ($9.00 a month) | 500 on Standard ($99.00 a month) | Oct 9, 2026 |
| Scrape.do | 10 on Hobby ($29.00 a month) | 200 on Advanced ($699.00 a month) | Oct 9, 2026 |
| ScrapeGraphAI | 3 on Starter ($20.00 a month) | 50 on Pro ($500.00 a month) | Oct 9, 2026 |
| ScrapeOps | 1 on 25,000 credits ($9.00 a month) | 200 on 10,000,000 credits ($699.00 a month) | Oct 9, 2026 |
| ScrapeOwl | 10 on Bootstrap ($29.00 a month) | 50 on Business ($249.00 a month) | Oct 9, 2026 |
| ScraperAPI | 20 on Hobby ($49.00 a month) | 500 on Advanced ($1,975.00 a month) | Oct 9, 2026 |
| Scrapfly | 5 on Discovery ($30.00 a month) | 100 on Enterprise ($500.00 a month) | Oct 9, 2026 |
| ScrapingBee | 50 on Freelance ($49.00 a month) | 800 on Enterprise 69 ($3,599.00 a month) | Oct 7, 2026 |
| ScrapingBot | 10 on Starter ($49.99 a month) | 200 on Business+ ($999.99 a month) | Oct 9, 2026 |
| ScrapingDog | 5 on Lite ($40.00 a month) | 2200 on Nova Pro ($30,000.00 a month) | Oct 9, 2026 |
| SERPHouse | 60 on Basic ($29.99 a month) | the same on every plan that publishes it | Oct 9, 2026 |
| Spider.cloud | 2 on Unlimited 2 concurrency ($40.00 a month) | 1538 on Unlimited 1538 concurrency ($19,994.00 a month) | Oct 9, 2026 |
| Steel.dev | 100 on Scale ($250.00 a month) | the same on every plan that publishes it | Oct 7, 2026 |
| WebScraping.AI | 10 on Personal ($29.00 a month) | 50 on Startup ($249.00 a month) | Oct 9, 2026 |
| WebScrapingAPI | 20 on Starter ($19.00 a month) | 500 on Pro ($599.00 a month) | Oct 9, 2026 |
| ZenRows | 20 on Build B1 ($19.00 a month) | 400 on Scale S5 ($2,999.00 a month) | Oct 9, 2026 |
In more detail
The vendor examples below were read on the dates given in the text; the table above holds each vendor's current figure.
Concurrency changes a bill when the required parallelism forces a higher plan even though a lower plan contains enough credits. It also determines whether the buyer must operate a queue. If excess requests receive an error instead of waiting, retrying them without a client-side limit can create more attempts or missed work. This is an operating and plan-fit issue, not a claim about vendor speed. (ScraperAPI pricing and Spider.cloud pricing, checked 2026-07-26.)
ScraperAPI publishes concurrent-request limits that rise from twenty on Hobby to five hundred on Advanced. Requests above the account limit receive HTTP 429 and are not queued. The vendor recommends controlling concurrency on the client side, so the buyer must keep no more than the plan limit in flight. (ScraperAPI pricing and ScraperAPI concurrency FAQ, checked 2026-07-26.)
ScrapingBee publishes fifty concurrent requests on Freelance, one hundred on Startup, two hundred on Business, and four hundred on Business+. Those limits sit beside the credit allowances, which means capacity and parallelism are separate plan dimensions even when both rise together. (ScrapingBee pricing, checked 2026-07-26.)
Firecrawl publishes API concurrency from five on Hobby to one hundred on Growth and Scale. Its Browser Sandbox has a different rule: the documentation currently allows up to twenty concurrent browser sessions on all plans. A buyer using both endpoints cannot treat the plan’s general API concurrency as the browser-session limit. (Firecrawl pricing and Firecrawl Browser Sandbox, checked 2026-07-26.)
Spider.cloud’s Unlimited offer makes concurrency the purchased plan dimension while allowing unlimited requests within it. Requests above the purchased concurrency receive HTTP 429 rather than queueing and are not charged. Its pay-as-you-go offer instead publishes request-rate capacity, illustrating why rate and concurrency must be recorded separately. (Spider.cloud pricing, checked 2026-07-26.)
Before signing up, identify the exact limited object: request, connection, session, crawl, or asynchronous job. Ask whether excess work queues or errors, whether endpoint pools are shared, and whether retries occupy another slot. Model the longest normal request duration and use a client-side semaphore. Compare plans on both allowance and required parallelism. (ScraperAPI concurrency FAQ, Firecrawl Browser Sandbox, and Spider.cloud pricing, checked 2026-07-26.)