-
Notifications
You must be signed in to change notification settings - Fork 4.7k
browserbase alternatives
"Browserbase alternatives" is two different searches wearing one query, and the honest first step is deciding which one is yours. Either you need what Browserbase actually sells - managed browser fleets in the cloud - and want a different vendor or a self-hosted version of the same thing. Or you arrived here from the top of the funnel, wanting "an agent that browses", found that the famous name in the space is an infrastructure company, and what you actually need is an agent, at which point the infrastructure bill is optional.
This page answers both, with the disclosure up front: it lives on the wiki of invisible_playwright_mcp, an open-source local agent that appears in the second half. Every claim about Browserbase and the other vendors below was read from their own sites, pricing pages and repositories on 2026-09-03.
Fairness before alternatives. Browserbase is the mindshare leader of the browser infrastructure category and earned it. The pricing is legible: a free tier (3 concurrent browsers, sessions capped at 15 minutes), a $20/mo Developer tier with 100 browser hours then $0.12/hr, a $99/mo Startup tier with 500 hours, and custom above that. Its homepage claims "10,000+ companies building beyond the API"; that is the company's own number, but the ecosystem gravity around it is checkable another way: it open-sourced Stagehand (MIT, ~24.1k stars), an SDK for browser agents that people use who never buy the cloud, and it gives away Director, which turns plain-English instructions into working agents and exports real Stagehand code. A vendor whose free, open layers are this good is not one you leave casually.
So the reasons to look elsewhere are specific, not general: you want open source you can self-host and audit; you want to own the metered costs at scale; you want a different point on the price curve; or, second half of the page, you never needed managed fleets at all.
If concurrent browsers in the cloud are genuinely your requirement - a product feature, CI, scale beyond one machine - these are the real alternatives, one line of facts each, read from their own pages on 2026-09-03.
| Alternative | Open? | The one-line honest version |
|---|---|---|
| Steel | Apache-2.0, ~7.6k stars | The credible open-source answer: a browser API you self-host with one docker run, Chrome over CDP, Puppeteer/Playwright/Selenium supported, plus a paid cloud if you want one |
| BrowserStation | MIT, ~166 stars | New and small, but architecturally serious: Kubernetes plus Ray, Chrome sidecars exposing CDP, self-described as an open-source alternative to Browserbase |
| Browserless | SSPL-1.0 OR commercial, ~13.7k stars | The veteran (repo since 2017), Docker-deployable, but the SSPL is source-available, not OSI-approved open source; read the license before you build on it |
| Hyperbrowser | SaaS | Hosted sessions plus scraping and crawl APIs, reachable over Playwright, Puppeteer and CDP |
| Anchor Browser | SaaS | Sells its own Chromium fork it calls "humanized chromium", in credit tiers from free to $2,000/mo; the fork's properties are Anchor's claims, untested by us |
| Kernel | SaaS | Chromium sessions over Playwright, CDP or WebDriver BiDi, and a hosted MCP server so tool-calling assistants can drive it directly |
| Cloudflare Browser Run | Big-cloud SaaS | Headless Chrome on Cloudflare's network, driven over Puppeteer, Playwright, CDP or Stagehand; the safe-default choice if you are already on Workers |
Three notes the table cannot hold. First, Steel is the head-to-head comparison most people searching this query actually want: open code, Apache license, self-host by default, cloud optional - the inverse of Browserbase's shape. Second, if your reason for leaving is price at scale, self-hosting moves the spend from browser-hours to your own compute and the engineering time to run it; that trade has a break-even, not a winner, and what a self-hosted AI agent costs to run puts one side of it in megabytes rather than adjectives. Third, every entry in the table, Browserbase included, runs Chromium or speaks its protocol, so switching vendors changes your operations and your bill, not what a website sees when it inspects the browser. The category-wide explainer covers that through-line: cloud browser infrastructure for AI agents.
Now the other search. If you are one person or one team wanting tasks done in a browser - not building a product that hosts browsers for others - then managed fleets are an answer to a question you do not have. An agent that runs on your own machine does the browsing where you are, and the infrastructure line item is zero because there is no infrastructure vendor.
The open-source agent field is surveyed properly on open-source AI browser agents; the short version relevant here:
- browser-use (MIT, the most-adopted agent framework) runs locally against a Chromium-family browser, and its company also now sells hosted infrastructure if you later grow into wanting it.
- Stagehand itself is MIT, which is worth restating in a page about its maker: the SDK layer is not what you pay Browserbase for; the managed fleet is. If your usage of Browserbase is "I like Stagehand", the code is open. How Stagehand compares to the agent frameworks is its own page: Stagehand vs browser-use.
- invisible_playwright_mcp (MIT, ~30k stars) - ours, disclosure above. It runs locally and ships its own browser: a Firefox patched at the C++ level rather than a stock Chromium build, driven through an MCP server from an assistant you already run (Claude Code, Claude Desktop, Cursor) or from its own UI with an OpenRouter key. The costs are model tokens and nothing else.
And the concessions, because this half is where our interest lives: a local agent gives you no fleet, no concurrency beyond your machine, no session cloud, no CI story, and invisible_playwright_mcp specifically is Windows and Linux only. A hardened browser changes what a page reads from the browser; it does not change your IP, your pacing, or a site's limits, and nothing here promises otherwise - why agents get blocked draws that boundary in full. If your workload is a product's workload, half one is your half.
- You operate browser fleets for a product or pipeline, want managed: stay on Browserbase, or price Hyperbrowser, Kernel, Anchor and Cloudflare against your usage shape.
- Same requirement, but open source and self-hosted: Steel first; BrowserStation if you are Kubernetes-native and early-adopter tolerant; Browserless only after reading the SSPL.
- You want an agent for your own tasks on your own machine: the agent field above; the infra bill disappears. invisible_playwright_mcp is our entry in it, conflict of interest noted.
- You mainly want Stagehand: it is MIT; you can use it without the cloud, and Stagehand vs browser-use maps that choice.
What is the best open-source alternative to Browserbase? Steel (Apache-2.0) is the mature one: self-hosted browser API, Chrome over CDP, one Docker command to try. BrowserStation (MIT) is the newer Kubernetes-native option. Browserless self-hosts but is SSPL-licensed, which is not OSI open source.
Is Browserbase free? It has a real free tier: 3 concurrent browsers, 1 browser hour, 15-minute session cap (read 2026-09-03). Paid starts at $20/mo. Stagehand and Director are free layers around the paid cloud.
Can I self-host Browserbase? No; the cloud is the product. Stagehand, its MIT SDK, runs wherever you want. For the managed fleet itself, the self-host equivalents are Steel and BrowserStation.
Do I need Browserbase to run an AI browser agent? No. Every major open-source agent runs on your own machine. The cloud earns its cost at fleet scale, in CI, or inside products; for personal and single-machine use, a local agent (invisible_playwright_mcp among them, disclosure above) has no infra cost at all.
Is Browserbase open source? The platform is not; Stagehand (MIT) and Director's exported code are the open layers. That split - open SDK, paid cloud - is the category's standard shape, not a Browserbase quirk.
See also: Cloud browser infrastructure for AI agents for the whole category explained, Stagehand vs browser-use for the SDK-level choice, open-source AI browser agents for the agent field, and browser-use alternatives for the same exercise applied to the biggest agent framework.
All retrieved 2026-09-03; prices, limits and star counts were read from each vendor's own page on that date and will drift.
- Browserbase, its pricing page, its docs and Director.
- browserbase/stagehand.
- steel-dev/steel-browser.
- ReinforceNow/browserstation.
- browserless/browserless.
- Hyperbrowser docs.
- Anchor Browser.
- Kernel.
- Cloudflare Browser Run.
- browser-use.com, for its hosted infrastructure product.
Written while maintaining invisible_playwright_mcp, which competes with exactly one half of this page. That is why the other half opens with what Browserbase gets right and sends you to Steel, not to us.
- OpenAI Operator alternatives
- Open-source Operator-style agents
- Is OpenAI Operator still available?
- OpenAI Operator vs Claude computer use
- browser-use alternatives
- Choosing an AI browser agent
- Open-source AI browser agents
- Open-source computer-use agents
- What is an AI web agent?
- AI browser agents vs traditional scraping
- Cloud browser infrastructure for AI agents, explained
- Browserbase alternatives
- Firecrawl vs an AI browser agent
- Skyvern alternatives
- Stagehand vs browser-use
- Project Mariner is gone: what replaced it
- Manus alternatives
- Gemini computer use vs Claude computer use
- invisible_playwright_mcp, reviewed honestly by its own wiki
- AI browser vs AI browser agent: which one do you want?
- AI browser agent vs RPA: which one fits the job
- AI browser agent vs n8n, Zapier and Make
- Vercel agent-browser alternatives, compared honestly
- What is an agentic browser? Definition and the two kinds
- Open-source agentic browsers: the three layers, compared
- Choosing an MCP server for browser automation: four axes
- Stealth MCP servers compared: Camoufox, nodriver, Patchright
- Playwright MCP alternatives, and the three you don't need
- Autonomous browser agents: the four rungs of autonomy
- What is actually free in the AI browser agent stack
- browser-use on GitHub: what the repo actually gives you
- Playwright MCP vs Chrome DevTools MCP: different jobs
- How to choose among MCP servers: a map by category
- Which MCP servers are worth adding to Claude Code
- MCP on GitHub: finding servers and judging them fast
- MCP vs an API: the decision, and what the wrapper costs
- MCP alternatives: when the protocol is the wrong shape
- Why does my AI agent get blocked?
- The timing signal AI agents give off
- Agent retry loops trip rate limits, not fingerprints
- Claude computer use detected as a bot
- browser-use getting blocked: what you can and cannot change
- Playwright MCP session blocked: four causes, four fixes
- Playwright MCP and captchas: what actually gets you past
- Cloudflare and a browser MCP server: what is being read
- Can an AI agent solve a captcha? The honest answer
- Getting an AI agent to fill out forms
- Which model to use with invisible_playwright_mcp
- Browser problem or model problem?
- Running invisible_playwright_mcp's browser from Claude Code
- Extracting data to a CSV with an AI agent
- Monitoring a page for changes with an AI agent
- Running invisible_playwright_mcp's browser from Claude Desktop
- Running invisible_playwright_mcp's browser from Cursor
- Using an AI agent to hunt for apartments
- Getting website data into Google Sheets with an AI agent
- Using an AI agent to download invoices from portals
- AI agents for web research
- Using an AI agent to test your own website
- Running invisible_playwright_mcp's browser from Cline
- Posting to social media with an AI agent
- Posting to Facebook with an AI agent
- Posting to Instagram with an AI agent
- Posting to X with an AI agent
- Automating LinkedIn posts: read this first
- Appointment bots: what they are and what an agent can legitimately do
- Track prices across sites with an AI agent
- Build a lead list with an AI browser agent
- Run an AI browser agent on a schedule
- AI browser agent with a local LLM: what changes
- Should you log your AI agent into your accounts?
- How to write a task an AI browser agent can follow
- Move data between two web apps with an AI agent
- The MCP server
- How the tools are shaped, and why
- Playwright MCP vs the Playwright CLI: which fits when
- Playwright MCP: browser is already in use, and the fix
- Playwright MCP best practices: four decisions that matter
- Playwright MCP with a proxy, and the three leaks it leaves
- A browser MCP server in GitHub Copilot: setup and limits
- Using a browser MCP server for web scraping: the pattern
- Which LLM for browser automation: the four properties
- How to build a browser agent, and what to take instead
- Getting an AI agent to log into a website: three routes
- MCP tools, resources and prompts: who controls each
- How many MCP tools is too many? The context arithmetic
- How to build an MCP server: the decisions, not the scaffold
- Local or remote MCP server: what changes, and what does not
- Writing an MCP client in Python: the thirty-line version
- Self-hosted AI agent: what one actually costs to run
- How long an AI browser agent takes per step, measured
- Text, HTML, snapshot or screenshot: what the agent should read
- Giving an AI browser agent a stopping condition
- Keeping an AI browser agent out of destructive actions
- Why did the AI agent click the wrong thing
- When the page changes under the AI agent
- Running one AI agent task across a list of sites
- Seeing a page as it appears in another country
- Getting data out of a dashboard with no export button
- Two browsers in one session: main and support
- Finding the dead links on a site with an AI agent
- Filling a CRM record from a company's website
- One form submission per spreadsheet row, with an AI agent
- Dated screenshots of a page as evidence
- Checking order and delivery status with an AI agent
- Reading a PDF that opens inside the browser
- Summarising a long page or thread with an AI agent
- Collecting every image on a page with its caption
- Collecting event and course listings with an AI agent
- Cancelling a subscription with an AI agent
- What an AI agent can and cannot do inside an iframe
- Shadow DOM and an AI agent: you can click it, you cannot read it
- Running invisible_playwright_mcp's browser from Codex
- What a page snapshot costs, per control
- Native selects and the ones that only look like selects
- Running invisible_playwright_mcp's browser from Gemini CLI
- Clicking by selector or by coordinates
- How long the agent waits before it gives up
- What a second browser costs
- Uploading a file with an AI agent, and why this one cannot
- Watching the agent work, and when it is worth it
- When not to use an AI browser agent
- Agent or script: deciding once instead of every time
- Using the keyboard instead of the mouse
- Secrets in an agent task: where they end up
- What an agent run should log
- Deduplicating what an AI agent collects
- Normalising values across sites
- Validating an AI agent's output
- Reading a table with an AI agent
- Driving a site's own search and filters
- The task works headed and fails headless