How Does Browserless Pricing Actually Work?
Browserless pricing is billed in units, not minutes: one unit buys up to 30 seconds of browser time on a single connection, so a 5-minute job costs 10 units. Your bill is the monthly unit allowance plus any overage, any proxy traffic converted into units, and any reconnect that opens a fresh connection. The plan name matters far less than the shape of your jobs.
📌 TL;DR Executive Summary
- Core Takeaway: One unit equals up to 30 seconds on one browser connection, rounded up. A 5-minute session is 10 units, and a reconnect starts the meter again from zero.
- Key Risk/Challenge: Proxy traffic and retries dominate real invoices — residential traffic costs 6 units per MB against 2 for datacenter, and overage runs $0.0020 per unit on Prototyping down to $0.0015 on Scale.
- Recommended Solution: Estimate in units from your own job size (seconds ÷ 30, times volume, plus retries and megabytes), then pick the tier whose session cap and allowance both fit.
Browserless Pricing, Defined
Browserless runs Chromium on its own infrastructure and lets your code drive it over a network endpoint. You send a Playwright, Puppeteer or REST request, a browser starts on their side, your script runs, and the browser closes. You never patch Chromium or maintain a container fleet yourself.
Because you don’t rent a machine, there’s nothing to bill by the hour. The unit is the billing atom, and that single design choice explains almost every surprise on a first invoice.
The unit, precisely
A unit is a block of up to 30 seconds of browser time on one connection. Anything past 30 seconds adds another unit, so 300 seconds bills as 10 units. Short jobs round up: a 4-second request still costs a whole unit, which makes thousands of tiny requests expensive per second worked.
What the monthly fee actually covers
A plan sells you a unit allowance plus operational ceilings: how long one session may run, how long persisted sessions and replays are kept, how far back logs are searchable. The allowance is what you spend; the ceilings decide whether your workload can run on that tier at all.
- Free: 1,000 units per month, a 2-minute session cap, 1 day of persisted sessions and replays, and two concurrent browsers as listed on the vendor page.
- Prototyping to Scale: larger allowances, 15- to 60-minute session caps, persistence rising from 7 to 90 days, and log search windows from 24 hours to 7 days.
- Enterprise: custom quote with private deployment or a licensed self-hosted install, plus SSO and a dedicated account manager.
How the Unit Model Works Under the Hood
Every connection starts a browser session on their side. Browserless meters wall-clock time from connection open to connection close, converts it into 30-second blocks and deducts them from your allowance. Proxy transfer is converted into units at fixed multipliers rather than billed in gigabytes.
The meter follows the clock, not your script’s productive work. A job that waits 40 seconds for a selector still burns 40 seconds of billable time, and a job that crashes after 20 seconds still paid for its block.
| Plan | Units per month | Max session | Overage per unit | Persisted sessions |
|---|---|---|---|---|
| Free | 1,000 | 2 min | No published rate | 1 day |
| Prototyping | 20,000 | 15 min | $0.0020 | 7 days |
| Starter | 180,000 | 30 min | $0.0017 | 30 days |
| Scale | 500,000 | 60 min | $0.0015 | 90 days |
| Enterprise | Custom | Custom | Custom | Custom |
The overage rate is the number to watch. Prototyping overage is about a third higher per unit than Scale: 20,000 extra units cost $40.00 on Prototyping and $30.00 on Scale. Same units, different rate, which is why heavy users move up a tier even when they could absorb the overflow.
Third-party trackers list Prototyping, Starter and Scale at $25, $140 and $350 per month billed annually: $300, $1,680 and $4,200 a year. No confirmed month-to-month list price is published, so treat those figures as annual commitments. Prices move, so check the official Browserless pricing page before you budget a quarter on it.
Switching is cheap by design: existing Playwright and Puppeteer scripts connect by changing the endpoint, not the code. The same pattern shows up in any walkthrough of how to call the Browserless API from Python or Node — one URL swap and the script runs on their machines.
Where the Bill Really Comes From: Proxies, Reconnects and Timeouts
Unit allowances look generous until you push traffic through the multipliers. Three things inflate an invoice faster than extra jobs.
Proxy traffic is priced per megabyte in units
Residential proxy transfer costs 6 units per MB; datacenter transfer costs 2 units per MB. At Prototyping’s overage rate, one megabyte of residential traffic is 6 units, about $0.012, and a gigabyte is 6,000 units, roughly $12. A job that pulls images, fonts and tracking pixels through a residential proxy can spend more units on bandwidth than on browser time.
Third-party estimates put residential traffic near $12.29 per GB and datacenter near $4.10 per GB once units are converted to dollars. The ordering matters more than the decimal: residential runs about three times datacenter, and both are metered on bytes actually transferred. Measure megabytes per job before you sign, then compare residential proxy prices against that number rather than a headline rate.
Every reconnect is a new connection
A dropped socket, a retry, a job that restarts after a failure — each one opens a new browser connection and bills a fresh unit. The Session API keeps cookies and localStorage alive across a disconnect, but restoring state doesn’t refund the block. If your suite retries twice on failure, budget three units for a one-unit job.
Session caps split long jobs into extra runs
Free caps a session at 2 minutes, Prototyping at 15. A 40-minute scrape can’t run as one session on Prototyping; you split it, and every segment pays rounding at both ends. Moving to Scale’s 60-minute cap isn’t only a bigger allowance — it changes the shape of the meter for long-running work.
Who Browserless Pricing Affects, and Who It Doesn’t
Metered browser pricing is built for machine-driven traffic: test suites on every commit, scheduled scrapes, checkout and uptime checks, and increasingly AI agents that call a browser as a tool. Browserless has added more API interfaces, including MCP and BrowserQL, aimed at AI-agent workflows as well as scraping.
For that work the unit model is a fair trade. You get managed Chromium, concurrency you can raise for a burst, and no container fleet to maintain in-house. You pay for seconds used, including seconds spent waiting and retrying.
It fits a different workload badly: a person signing into a client account, checking an order queue, posting an update. Those sessions have no script, no predictable duration and no endpoint. Metering them means paying by the minute for a human to read a dashboard.
How to Estimate Your Browserless Bill in Six Steps
Work in units the whole way and convert to dollars once, at the end. Mixing minutes and unit prices mid-estimate is how budgets drift.
- Time one real job. Measure end to end, including navigation and waits, and use the median run rather than the fastest.
- Round up to 30-second blocks. A 95-second job is 4 units, not 3.17. Keep the ceiling function.
- Multiply by monthly volume. Jobs per month × units per job gives your base load.
- Add a retry factor. If 8% of jobs retry once, add 0.08 × job count in units — reconnects bill in full.
- Add proxy units. Median MB per job × 6 for residential or × 2 for datacenter, times monthly jobs.
- Compare against the allowance and price the gap. Anything over the tier allowance bills at that tier’s overage rate.
| Line item | Working | Units per month |
|---|---|---|
| Browser time | 3,000 jobs × 90 s = 3 units each | 9,000 |
| Retries | 8% of 3,000 jobs × 1 unit | 240 |
| Residential proxy | 150 MB × 6 units per MB | 900 |
| Total | 10,140 |
That total sits comfortably inside Prototyping’s 20,000 units. Change one variable: at 5 minutes per job, 20,000 jobs need 200,000 units, which lands 20,000 units over Starter’s 180,000 allowance — $34 of overage at $0.0017. Run the same workload on Prototyping and 180,000 units of overage bill at $0.0020, or $360. The tier you pick changes the price of your overflow, not just your allowance.
Self-Hosting: What Browserless Pricing Doesn’t Include
The open Docker image — ghcr.io/browserless/chromium or /multi — is free under SSPL-1.0 and covers Puppeteer, Playwright and the core REST APIs. You run it in your own VPC, on-prem or air-gapped, swap the endpoint and keep your code unchanged. For an open-source project, the licence line is zero.
It isn’t free for everyone. Closed-source commercial products, and closed-source CI pipelines running Browserless, need a paid commercial licence. The Enterprise self-hosted image adds BrowserQL, session recording, RBAC, OpenTelemetry, GPU infrastructure and SSO. Read the self-hosted licensing details before assuming the image replaces a cloud plan.
The rest of total cost of ownership is the part budgets omit: containers, a load balancer, Chromium updates, memory headroom and someone on call. Vendor deployment guidance puts one deployment at roughly 5-10 concurrent light sessions, 10-20 medium and 20-50 heavy, which tells you how many containers a real workload needs. If you’re weighing hosted endpoints at all, this breakdown of Browserless vs Puppeteer cloud options covers the operational trade-offs in more depth.
The Plan Constraints That Decide Your Tier, Not the Price
Unit allowances are the visible number, but four ceilings decide whether a tier can run your workload at all. Check them in this order before you compare overage rates.
- Max session time. 2 minutes on Free, 15 on Prototyping, 30 on Starter, 60 on Scale. A job that needs 20 minutes cannot run on Prototyping at any unit volume.
- Concurrency. Bursty test suites hit simultaneous-browser limits before they hit unit limits, and third-party checks report that paying annually raises the concurrency ceiling on the same plan. Confirm the live figure for the period you’re buying.
- Session persistence and replays. 1 day on Free up to 90 days on Scale. Debugging a flaky job two weeks later needs the storage window, not more units.
- Log search. 8 hours on Free, 24 hours on Prototyping, 3 days on Starter, 7 days on Scale — the window in which you can find out why a job burned its allowance.
Enterprise is a different purchase rather than a bigger bucket: private deployment, GPU infrastructure, SSO, custom proxy geotargeting and a dedicated account manager. And sticker prices across adjacent tools don’t line up — Apify starts at $39/month and BrowserStack runs from $12.50 to $449/month, each metering something different — so convert any candidate to cost per job. Before you turn a monthly estimate into an annual contract, read a side-by-side of Browserless alternatives compared.
Where Send.win Fits: The Unmetered Half of the Work
Browserless prices machine hours. The sessions your team runs by hand — signing into a client’s storefront, checking an ad account, clearing a support inbox — don’t belong on a metered browser API. They belong in a browser where each account keeps its own cookies, storage and fingerprint settings, and nobody is charged by the minute.
How Send.win Helps With Browserless Pricing
Send.win is a browser profile manager for professionals and teams: every account gets its own isolated browser profile.
- Isolated profiles – separate cookies, storage and sessions per profile
- Fingerprint protection – each profile keeps its own consistent browser fingerprint, so profiles never share one browser identity
- Desktop app + cloud browser – native app for Windows, macOS and Linux, or run profiles in the cloud with no install
- Residential proxies included – on Pro and Team, with timezone and locale matched to the proxy
- Team sharing – share a signed-in profile with a teammate without sharing the password
Try the free cloud demo — no install — or download the desktop app. Paid plans start at $19/month, and your first card purchase is covered by a 7-day money-back guarantee (see pricing).
Send.win is a browser profile manager built for that. Each isolated profile carries its own cookies, storage, sessions, fingerprint settings and proxy, so one client’s logins stay separate from the next. There are two ways to run it: Sendwin Browser, a native desktop app for Windows 10/11, macOS 12+ and Linux, or the cloud browser running on Send.win’s server in Helsinki from any device. Timezone, locale, WebRTC and geolocation follow the proxy’s exit IP automatically, and Pro and Team include residential proxies with metered bandwidth.
Collaboration is the other half. Share a profile with a paid teammate and it opens already signed in, so no password changes hands, and live cloud sessions can be shared too. Pricing is by profile count and seats rather than browser-seconds: Pro is $19/mo for 150 profiles, 5 GB/month of proxy bandwidth and 6 seats; Team is $49/mo for 500 profiles, 20 GB/month and 16 seats.
Send.win is a hands-on tool: you open a profile and work in it yourself, and nothing in the product is designed to be scripted or driven programmatically. That keeps the split clean — machine-driven jobs run on a metered service built for them, while human account work runs in isolated profiles. On either side, run only accounts you own or are authorised to manage, and follow each platform’s terms.
🏆 Send.win Verdict
Browserless pricing is honest once you understand the unit — it’s the per-megabyte proxy traffic and per-reconnect billing that catch people out. But a metered browser API is the wrong tool for the part of your month that involves a person logging into a client account. Send.win covers that half without a meter: isolated profiles per client, shared logins that open already signed in, and a flat per-profile price, while machine-driven, scripted workloads stay on a metered service built for them.
See Send.win plans — Pro from $19/mo with 5 GB of proxy bandwidth, Team from $49/mo, backed by a 7-day money-back guarantee on your first card purchase, or try the free cloud demo first.
Frequently Asked Questions
What is a Browserless unit and how long is it?
One unit is up to 30 seconds of browser time on a single connection, so a 5-minute session is 10 units. Jobs shorter than 30 seconds still cost one full unit. That rounding is why thousands of tiny requests cost more per second of real work than a few longer jobs.
How much does Browserless cost per month?
Third-party trackers list Prototyping, Starter and Scale at $25, $140 and $350 per month billed annually, which works out to $300, $1,680 and $4,200 a year. No confirmed month-to-month list price is published. Enterprise is quoted privately, and the Free tier costs nothing within its caps.
Does Browserless have a free plan?
Yes: 1,000 units a month, a 2-minute session cap, 1 day of persisted sessions and replays, and two concurrent browsers per the vendor page. That is roughly 100 five-minute sessions. It’s enough to confirm your script connects, not enough to benchmark real throughput.
How much do residential proxies cost on Browserless?
Residential transfer costs 6 units per MB and datacenter transfer 2 units per MB. Third-party estimates put those at roughly $12.29 per GB and $4.10 per GB once converted to dollars. Both are metered on actual bytes moved, so image-heavy pages cost more than text pages.
What happens if I go over my unit allowance?
Overage is billed per unit at your tier’s rate: $0.0020 on Prototyping, $0.0017 on Starter and $0.0015 on Scale. Free has no published overage rate, so treat its 1,000 units as a ceiling rather than a soft limit. Because the rate falls as you move up, 20,000 extra units cost $40.00 on Prototyping but $30.00 on Scale.
Is the $25 Prototyping price monthly or annual?
Annual. It’s the monthly equivalent of a $300 yearly commitment, not a cancel-anytime monthly rate. Treat any comparison that puts $25 against another tool’s monthly price as apples-to-oranges until both are converted to the same billing period.
Is self-hosting Browserless free?
The open Docker image is free under SSPL-1.0 for Puppeteer, Playwright and the core REST APIs. Closed-source commercial products and closed-source CI need a paid commercial licence, and the Enterprise image with BrowserQL, session recording, RBAC and SSO is licensed separately. Hosting costs — containers, load balancer, updates, on-call time — are yours either way.
Why is my bill higher than my unit estimate?
The usual causes are proxy megabytes, retries and idle waiting. Residential transfer at 6 units per MB can outspend browser time on media-heavy jobs, and every reconnect bills a fresh unit. Log seconds, megabytes and retry count per job for a week; the gap between estimate and invoice usually shows up in one of those three columns.