One browser platform. Five ways to drive it.
BrowserQL, Puppeteer, Playwright, REST, or MCP. Pick the interface that fits the task. They all run on the same managed, stealth-ready browsers, so your automation scales without the infrastructure burden.
Free plan · No credit card · Cloud or self-hosted
The runtime under everything
Every Browserless API, whether GraphQL, WebSocket, HTTP, or MCP, connects to the same managed browser fleet. We call it Browsers as a Service: a pool of Chromium, Chrome, Firefox, and WebKit instances kept patched, isolated, and ready to launch in any of three regions.
Five interfaces
Runtime capabilities
Managed Chromium, Chrome, Firefox, and WebKit fleet
Stealth, CAPTCHA solving, residential proxies, persistent sessions, and LiveURL are runtime features, not API features. Turn them on per request, with a URL flag or a single mutation, and they work the same whether you're sending a GraphQL mutation, a curl request, or a Puppeteer connect.
One runtime. Different ways to talk to it. Same reliability across all of them.
Five ways to drive a browser
Same runtime. Different control plane. Pick the one that matches your stack and how much control you want.
Best for: Bypassing sophisticated bot detection and CAPTCHAs at scale.
- Declarative mutations: goto, click, type, solve
- Built-in stealth, CAPTCHA solving, and human-like behavior
- Web IDE for building queries before you ship them
Best for: Existing Puppeteer/Playwright code that needs to run somewhere reliable.
- Drop-in WebSocket endpoint, no rewrites
- Persistent sessions, LiveURL handoff, regional load balancing
- Stealth and proxy options as URL flags
Best for: Screenshots, PDFs, content extraction, and one-off scrapes.
- 13 endpoints including /smart-scrape, /screenshot, /pdf, /search, /crawl
- No browser libraries required, works from any language
- Cascading fallbacks built into /smart-scrape
Best for: AI agents, Claude/Cursor/Windsurf workflows, and LLM automation.
- Hosted MCP server with Browser Agent + 8 REST tools
- Stateful sessions that survive across tool calls
- Integrations: Browser Use, Stagehand, LangChain, Vercel AI SDK
Best for: Data sovereignty, air-gapped environments, and regulated industries.
- Docker image: Chrome, Chromium, Firefox, WebKit, Edge
- Open-source or licensed Enterprise builds
- OpenTelemetry, NGINX load balancing, custom proxies
Not sure where to start?
Pick the task. We'll point you at the right surface.
What you get on every surface
These aren't features of one API. They're features of the managed runtime, and they work the same across BrowserQL, BaaS, REST, and MCP. Self-hosted deployments bring their own proxies and load balancing.
Stealth & CAPTCHA solving
Hide automation signals, solve reCAPTCHA and Cloudflare automatically.
Residential proxies
Route through real residential IPs to avoid blocks, opt in per request.
Persistent sessions
Reconnect to the same browser across requests, and keep cookies and local storage alive between sessions with authenticated profiles.
LiveURL handoff
Hand a session to a human for verification or interaction, then resume automation.
Regional endpoints
San Francisco, London, Amsterdam. Pick the closest, or distribute across all three.
SOC 2 + 99.9% uptime
Production-grade reliability with automatic failover and SOC 2 Type II controls.
Built for the AI era
AI agents need browsers that don't get blocked, don't lose state between tool calls, and don't crash under load. Browserless gives your agent that browser.
The hosted Browserless MCP Server drops into Claude Desktop, Cursor, VS Code, and Windsurf with one config block. Nothing to install. Or use Browserless directly inside Browser Use, Stagehand, LangChain, Vercel AI SDK, or the Claude Agent SDK. Every framework that matters has a path through.
See AI integrations{
"mcpServers": {
"browserless": {
"type": "http",
"url": "https://mcp.browserless.io/mcp",
"headers": {
"Authorization": "Bearer YOUR_API_TOKEN_HERE"
}
}
}
}docker run --rm -p 3000:3000 \
-e "TOKEN=YOUR_TOKEN" \
ghcr.io/browserless/chromiumOr run it yourself
Same APIs. Your infrastructure. The open-source Browserless image runs Puppeteer, Playwright, and the REST APIs inside your own VPC, on-prem, or in an air-gapped environment.
The licensed Enterprise image adds the rest of the platform: BrowserQL, session recording, the live debugger, and advanced webhooks, with OpenTelemetry and NGINX load balancing for production. All of it backed by the same SOC 2 Type II audited company.
See Self-HostedUsed by teams shipping browser automation at scale
“We were hosting our own Puppeteer-driven smoke testing service, which required specialized operational attention to maintain and scale. We began looking for third-party hosted solutions so that we could focus our attention on building and supporting our core products, and Browserless fit the bill.”
Christopher Zhen
Software Engineer, Samsara
“Browserless helped us focus on the problem we were trying to solve, and less on scaling an automation infrastructure. Browserless's developer focused approach has been a key to us bringing our product to market at the speed we were able to do so. Joel and team are some of the most customer-centric partners I've worked with.”
Scott Weinert
Co-Founder & CTO, Atomic
Pick your front door.
Start free in 60 seconds, or talk to an engineer about your workload. 1,000 free units/month · No credit card · Cancel anytime.