Browserbase rents you headless Chrome in the cloud. You connect Playwright, Puppeteer, Selenium or its own Stagehand framework to a remote browser, and Browserbase runs the fleet, the session recordings and the captcha solving. It is built for AI agents that log in, fill forms and click through portals, and a lot of teams also use it to scrape.
That second use is where the pricing page is easy to misread. The rates below were read from browserbase.com/pricing and the Browserbase docs on October 2, 2026. Prices are monthly in USD. Check the live page before you commit, since the plan lineup has changed more than once.
The four Browserbase plans and what each includes
| Plan | Per month | Browser hours | Proxy traffic | Concurrent browsers |
|---|---|---|---|---|
| Free | 0 | 1 hour, 15 minutes per session | none | 3 |
| Developer | 20 USD | 100, then 0.12 USD an hour | 1 GB, then 12 USD per GB | 25 |
| Startup | 99 USD | 500, then 0.10 USD an hour | 5 GB, then 10 USD per GB | 100 |
| Scale | custom | 500 and up, usage based | 5 GB and up, usage based | 250 and up |
Every paid plan adds captcha solving and basic stealth mode; Scale gets advanced stealth, verified agents, HIPAA with a BAA and SSO. Free keeps session data for 7 days, paid plans for 30. On top of browsers, each plan includes Fetch and Search API calls (1,000 each on Free and Developer, 10,000 Fetch calls on Startup) and a handful of managed agent runs. Model tokens through the Browserbase gateway bill at market price.
Browserbase bills three meters, and scraping mostly hits one
A page you scrape through a Browserbase session costs browser time plus the bytes that travel through its proxies. Browser time is metered per hour of session. Proxy traffic is metered per gigabyte, and a real browser downloads everything the page asks for: HTML, scripts, stylesheets, fonts, images and tracking calls.
Browserbase's own guide says a typical scrape runs in under two minutes and that 100 browser hours are roughly 3,000 page level tasks. That framing fits an agent filling in a form. A scraper that opens product pages in a warm session spends closer to 5 seconds per page, so 100 hours stretch to about 72,000 pages. Hours stop being the constraint almost immediately.
Gigabytes take over. The Developer plan's single included gigabyte covers about 3,300 pages if you block images and fonts and each page costs around 0.3 MB, or about 500 pages if you let a 2 MB retail page load in full. After that every gigabyte is 12 USD. This is the detail the plan cards do not spell out, and none of the third party reviews we read converts it into a per page figure.
Browserbase cost per 1,000 pages
The table prices 1,000 pages beyond the included allowance, at 5 seconds of browser time per page. Light means images, media and fonts blocked in your Playwright code, about 0.3 MB through the proxy. Full means the page loads as a visitor sees it, about 2 MB. Your own page weight will differ, so treat these as a method, then plug in your number.
| Way of fetching | Developer | Startup |
|---|---|---|
| Browser, no proxy | about 0.17 USD | about 0.14 USD |
| Browser with proxy, light page | about 3.77 USD | about 3.14 USD |
| Browser with proxy, full page | about 24.17 USD | about 20.14 USD |
| Fetch API, plain | 1.00 USD | 0.50 USD |
| Fetch API with proxies | 4.00 USD | 4.00 USD |
| Fetch Extract to JSON with proxies | 7.00 USD | 7.00 USD |
For comparison, one successful page on our plans is 0.98 USD per 1,000 on Starter, 0.60 on Growth and 0.33 on Scale, with rotating residential and datacenter proxies, browser rendering and field extraction in that price, whatever the page weighs. A failed page is not billed at all.
The Fetch API shortcut and its two limits
Fetch is the cheap way into Browserbase for plain pages: one URL in, raw HTML, Markdown or JSON out, no browser session. Two things in the docs matter before you design around it. Fetch does not execute JavaScript, so any page that builds its content in the browser comes back empty and has to go through a session after all. And redirects are off by default, so a product URL that 301s to its canonical returns the redirect, not the page, unless you set allowRedirects.
Fetch is also capped at 5 requests a second per project on every self serve plan, a 5 MB response and a 60 second timeout. Five a second is about 13 million calls in a month, which is plenty for most scrapers, but bursty jobs that want 50 pages at once will queue. Once proxies are on, the rate is 4 USD per 1,000 calls on both paid plans, so Startup's lower plain rate does not help you on protected sites.
Real monthly budgets on Browserbase and on flat per request pricing
Three workloads on sites that need proxies, priced on the cheapest Browserbase plan that fits, with light pages. The Fetch column assumes pages that do not need JavaScript; if the plan's included Fetch calls also count with proxies, take 4 to 40 USD off it.
| Pages per month | Browser sessions, light | Fetch with proxies | Flat per request |
|---|---|---|---|
| 50,000 | about 188 USD, Developer (15 GB) | about 220 USD, Developer | 49 USD, Starter |
| 250,000 | about 799 USD, Startup (75 GB) | about 1,099 USD, Startup | 149 USD, Growth |
| 1,000,000 | about 3,138 USD, Startup (300 GB) | about 4,099 USD, Startup | 499 USD, Scale |
With full page loads the proxy part of the session column grows six to seven times. The Scale plan is negotiated, and at a million pages a month Browserbase will almost certainly quote below these list rates, so ask before you rule it out. The shape of the gap holds, though: per gigabyte billing punishes heavy pages, and a flat price per successful page does not care what the page weighs.
Where Browserbase is the right buy
Agents that act, not just read. Logging in to a vendor portal, downloading statements, filing a form, clicking through a checkout. A scraping API returns a page; Browserbase gives your code a whole browser session, with session replays to debug what the agent saw. For that work it is the better tool and its pricing is fair.
Sites that do not need proxies. If the target does not block datacenter traffic, session time alone is about 0.14 to 0.17 USD per 1,000 pages, cheaper than any per request API including ours. Internal tools, partner sites and documentation fall here.
Teams already invested in Playwright. If you have a few hundred lines of page interaction you trust, pointing them at a remote browser is a one line change. If staffing that work is the actual bottleneck, it can be faster to hire an AI developer who has shipped browser agents before than to rebuild the scripts around a different API.
Where flat per request pricing wins
Reading protected pages at volume. Retail catalogs, travel fares, marketplaces and job boards sit behind Cloudflare, DataDome, Akamai or Imperva and need residential traffic on every request. That is precisely the case where Browserbase's per gigabyte proxy meter runs, and where one price per successful page is three to twenty times cheaper on the numbers above. Our Cloudflare scraping API page explains how the retries work.
Crawling whole sites. Browserbase has no crawler; you discover links and queue pages yourself. Our website crawler API takes a seed URL, a depth and include or exclude patterns and returns every page as JSON, billed per successful page like any other request.
Predictable budgets. A finance team can sign off on 149 USD for 250,000 pages. It is harder to sign off on a gigabyte meter that moves every time a retailer adds a hero video. If you are weighing other usage based vendors too, the same arithmetic for Apify and Bright Data is in our pricing series, and every plan of ours is on the pricing page.
Questions buyers ask
How much does Browserbase cost per month? Free is 0 with 1 browser hour and 3 concurrent browsers. Developer is 20 USD with 100 hours, 1 GB of proxies and 25 browsers. Startup is 99 USD with 500 hours, 5 GB of proxies and 100 browsers. Scale is quoted per customer, starting at 250 concurrent browsers.
How much is a Browserbase browser hour? 0.12 USD on Developer and 0.10 USD on Startup once the included 100 or 500 hours are used. Scale is usage based at a negotiated rate. Browser hours are rarely the expensive part of a scraping job; proxy gigabytes are.
How much do Browserbase proxies cost? 12 USD per GB on Developer after the included 1 GB, and 10 USD per GB on Startup after the included 5 GB. A browser pulls every asset through the proxy, so block images, media and fonts in your automation code to keep the meter down.
Is Browserbase free? There is a free plan with 1 browser hour a month, 15 minute sessions, 3 concurrent browsers, 1,000 Fetch and 1,000 Search calls, and no proxies or captcha solving. It is enough to test the integration, not to run a scraper.
Does the Browserbase Fetch API render JavaScript? No. Fetch returns the server response only, up to 5 MB, with a 60 second timeout and redirects off by default. Pages built in the browser need a full Browserbase session, which bills browser hours and proxy traffic instead.
Is Browserbase cheaper than a scraping API? For sites that do not block and for interactive agent work, often yes. For protected pages at volume, usually no: at about 0.3 MB per page through the proxy, Startup works out near 3.14 USD per 1,000 pages against 0.33 to 0.98 USD on our plans.