# Browserbase — Cloud headless browsers, search/fetch APIs and managed browser agents behind one API key. > Source: The Agents Index — https://theagentsindex.com/browserbase (structured, researched, re-verified) > Facts last verified: 2026-09-02 Browserbase is a cloud platform giving AI agents real headless browsers plus search, fetch, proxies, verified identity and a model gateway behind one API key. It also runs managed browser agents and ships the open-source Stagehand SDK. | Fact | Value | | --- | --- | | Website | https://www.browserbase.com | | Price | from $20/month | | Free tier | Yes | Free Plan $0/mo: 3 concurrent browsers, 1 browser hour, 3 agent runs, 1,000 search calls, 1,000 fetch calls, 15 minutes per session, 7 days data retention, $5 in model tokens, 1 project. No proxies and no captcha solving. No credit card needed. ## What it does **Writes code** · Partly — Generates Stagehand browser-automation scripts from a completed Agent run; not general application code. > Use the Optimize tool to turn a completed Agent run into a reusable Stagehand script. Optimize reads the run's steps and generates code that follows the same workflow. > — https://docs.browserbase.com/platform/agents/generating-scripts · Generate a script, opening paragraph **Debugs and fixes** · Partly — Documents a CI recipe where a coding agent inspects the live DOM on Browserbase and fixes broken selectors; the fix comes from Claude Code, not Browserbase. > This GitHub Action creates a Browserbase session, lets Claude Code inspect the live DOM, and opens a PR with fixed selectors. > — https://docs.browserbase.com/integrations/skills/github-action · Auto-fix selectors in CI/CD, Info callout **Writes tests** · No — Testing docs cover running end-to-end tests on hosted browsers; no page describes authoring tests. > Run end-to-end browser tests in consistent, isolated cloud environments. Browserbase gives you reliable infrastructure for automated testing workflows, whether you're using Stagehand or Playwright. > — https://docs.browserbase.com/use-cases/building-automated-tests · Automated testing, intro paragraph **Ships it** · Yes — Browser automations are deployed as hosted Functions, published and invoked from the CLI or API. > browse functions publish index.ts > — https://docs.browserbase.com/integrations/skills/browse-cli · Common workflows, Functions code block **Done-for-you service** · No — Self-serve software platform; consulting is offered only to architect hybrid on-prem deployments. > Browserbase is a cloud-first platform designed for scalability and reliability. > — https://www.browserbase.com/pricing · FAQ, "Can I run my browsers on prem?" ## How much it does unattended **Runs autonomously** · Yes — Managed Agents complete a prompted web task from one API call, with no human in the loop. > Run autonomous browser agents on Browserbase with one API call. > — https://docs.browserbase.com/platform/agents/overview.md · Agents overview, page description **Multi-agent** · Partly — Thousands of sessions run in parallel and Stagehand plugs into multi-agent frameworks; Browserbase itself does not orchestrate agents. > It gives developers Playwright-level control combined with AI primitives (act, extract, observe) and plugs into all major agent frameworks (LangChain, CrewAI, Mastra, etc.). > — https://www.browserbase.com/pricing · FAQ, "How is Browserbase related to Stagehand?" **Agent permissions** · Partly — Per-session domain allowlisting restricts navigation, but the feature is marked experimental and covers top-frame navigation only. > Allowed domains let you restrict which websites a session can navigate to. If you pass one or more domains, Browserbase blocks top-level page navigations to any domain not on the list. > — https://docs.browserbase.com/platform/browser/security/allowed-domains.md · Allowed Domains, "Restricting navigation" intro ## Which models it runs on **Claude** · Yes — Anthropic models are reachable through Model Gateway with only a Browserbase API key. > | Anthropic | `anthropic/claude-sonnet-4-6` | > — https://docs.browserbase.com/platform/model-gateway/overview.md · Supported providers table, Anthropic row **GPT** · Yes — OpenAI models are reachable through Model Gateway with only a Browserbase API key. > | OpenAI | `openai/gpt-5` | > — https://docs.browserbase.com/platform/model-gateway/overview.md · Supported providers table, OpenAI row **Open models** · No — The gateway's provider list is closed at OpenAI, Anthropic and Google; no open-weight model is named. > Need a provider that isn't listed? Reach out. Browserbase is happy to work with teams on additional model support. > — https://docs.browserbase.com/platform/model-gateway/overview.md · Supported providers table plus the tip immediately beneath it **Model choice** · Yes — You pick the model in Stagehand config, or omit it and let the gateway choose per call. > Switch models in your Stagehand config without touching credentials or account > — https://www.browserbase.com/models · "Experiment without overhead" feature block ## Where you use it **In your editor** · Partly — No editor extension; the CLI and MCP server are what run inside Cursor, Claude Code and other MCP clients. > Yes. Browserbase CLI is designed for coding agents: it is scriptable, discoverable, and works well inside terminal-based agent loops. In Claude Code, you can also run bb skills > — https://www.browserbase.com/browse-cli · FAQ, "Does Browserbase CLI work with Claude Code, Cursor, and Codex?" **On the command line** · Yes — Browse CLI fetches, searches, browses and manages sessions, contexts and Functions from the terminal. > Browserbase CLI gives agents one interface to fetch pages, search the web, control a real browser, manage sessions, persist browser state, and turn browser workflows into reusable automation. > — https://www.browserbase.com/browse-cli · FAQ, "What is Browserbase CLI used for?" **In your pipeline** · Yes — Documented GitHub Actions workflow creates a session, inspects the live DOM and opens a pull request. > This GitHub Action creates a Browserbase session, lets Claude Code inspect the live DOM, and opens a PR with fixed selectors. > — https://docs.browserbase.com/integrations/skills/github-action.md · Auto-fix selectors in CI/CD, info callout **In a browser** · Yes — Hosted dashboard with a playground, session inspector and Function deployment. > Create, test, and deploy browser agent Functions directly from the Browserbase dashboard > — https://docs.browserbase.com/llms.txt · Docs index entry, "Deploy from playground" description ## Whose machine it runs on **Self-hosted** · No — No on-prem deployment; private cloud and VPC connectivity are the enterprise substitutes. > Browserbase is a cloud-first platform designed for scalability and reliability. While we don't offer on-prem deployment, our cloud infrastructure provides: > — https://www.browserbase.com/pricing · FAQ, "Can I run my browsers on prem?" **Open source** · Partly — Stagehand SDK and the Browse CLI are open source; the hosted browser platform is not. > Stagehand is an open source SDK that uses AI to make your browser agents resilient, readable, and production-ready. > — https://www.browserbase.com/stagehand · Hero, "The SDK for browser agents" ## What it costs to run **How it meters** · Yes — Monthly plan fee plus pay-as-you-go overages metered per browser hour, per GB of proxy and per 1k API calls. > Each plan includes a monthly allocation of browser hours and proxy bandwidth (and more). Go beyond it and pay-as-you-go overage rates apply. No caps. No cut-offs. > — https://docs.browserbase.com/account/billing/plans.md · "What's included" section **Free tier** · Yes — Free plan with 3 concurrent browsers, 1 browser hour and 3 agent runs; no credit card and no sales call. > Start free, no credit card needed. Includes 3 concurrent sessions. > — https://docs.browserbase.com/account/billing/plans.md · Pick a plan card group, "Free $0" **API access** · Yes — REST API and Node/Python SDKs for sessions, agents, contexts, search, fetch and Functions on every plan. > Create, connect, and retrieve information from browser sessions. > — https://docs.browserbase.com/reference/api/overview.md · API reference overview, page description **MCP server** · Yes — Hosted MCP server at mcp.browserbase.com gives MCP clients browser automation via Stagehand. > This server provides cloud browser automation capabilities using Browserbase and Stagehand. It enables LLMs to interact with web pages, take screenshots, extract information > — https://www.browserbase.com/mcp · "What it does" section **Bring your own key** · Yes — You may bring your own model provider, your own proxies, and on Enterprise your own S3 storage. > Omit `model` to let Model Gateway select one for each `act`, `extract`, and `observe` call. See Stagehand model configuration to pin a model or bring your own provider. > — https://docs.browserbase.com/platform/model-gateway/overview.md · Setup section, note under the code sample ## Buying it for a team **A company can buy it** · Yes — Organizations with member invitations, projects, Stripe billing portal and invoice billing on Scale. > Invoice billing: Keep procurement happy and skip the credit card hassle. > — https://www.browserbase.com/enterprise · "What our Scale plan unlocks" list **Seat model** · No — Plans are priced per organization on capacity and usage; the plan tables carry no seat count or per-user price. > | | Free \$0 | Developer \$20 | Startup \$99 | Scale plan | | **Browser Hours** | 1 hr | 100 hrs | 500 hrs | Flexible | > — https://docs.browserbase.com/account/billing/plans.md · Browser allocations table — the enumerated plan dimensions contain no seat row **Pooled budget** · Yes — Allocations, retention and billing attach to the organization and its projects, not to individual members. > Settings that live on your Browserbase Organization, such as data retention, projects, and support tier. > — https://docs.browserbase.com/account/billing/plans.md · Account section intro, above the Account table **Admin controls** · Partly — Admin roles and project access govern the dashboard only; what a session may do is set per session in code, not by policy. > User roles only affect what permissions a user has within the dashboard itself. They don't change who can interact with sessions via the API. That's what your API key is used for. > — https://docs.browserbase.com/account/team/roles.md · User roles section, info callout under the permission table **Audit log** · Yes > Automate KYC checks across business registries, sanctions lists, license boards, and verification portals. Faster onboarding, fewer manual reviews, full audit trails. > — https://www.browserbase.com/use-cases · Vendor page, feature statement **Single sign-on** · Partly — SAML 2.0 SSO with JIT provisioning, but only on Enterprise/Scale plans and enabled by Browserbase support. > SSO is available only on Enterprise plans. Get in touch with the Browserbase team to enable SSO on your account. > — https://docs.browserbase.com/account/team/sso.md · Info callout under the intro ## What happens to your code **Opt out of training** · Yes — The privacy policy makes no-training the default, conditioned on affirmative consent. > We do not use your browser data to train any generative artificial intelligence models without your affirmative consent. > — https://www.browserbase.com/privacy-policy · How We Use Your Personal Data section **Data residency** · Yes — Region is chosen per session from us-west-2, us-east-1, eu-central-1 and ap-southeast-1, and residency is configurable for enterprise. > Available regions: `us-west-2` (Oregon), `us-east-1` (Virginia), `eu-central-1` (Frankfurt), and `ap-southeast-1` (Singapore). All sessions are created in `us-west-2` by default. > — https://docs.browserbase.com/optimizations/latency/multi-region.md · Note at the end of Browser regions **Getting out** · Partly — Artifacts are retrievable by API and replays downloadable as MP4, but retention is capped at 7 or 30 days by plan and monthly allocations do not roll over. > | **Data Retention** | 7 days | 7 days | 30 days | 30+ days | > — https://docs.browserbase.com/account/billing/plans.md · Account table, Data Retention row **Certifications** · Yes — SOC 2 Type II on every plan; HIPAA BAA and DPA on Scale only; third-party penetration testing. > SOC 2 Type II Certified; HIPAA compliant: Business Associate Agreements (BAAs) available; Comprehensive third-party penetration testing with successful results > — https://docs.browserbase.com/account/enterprise/security.md · "Enterprise-grade compliance" bullet list ## Provenance 1 of 44 sources behind the facts above are independent of the vendor. ## Coverage elsewhere Pages about this product that the vendor does not control. The label states how independent each one is. | Source | Kind | Read | | --- | --- | --- | | respan.ai | Independent source | 2026-08-17 | | skyvern.com | Independent source | 2026-08-17 | | tinyfish.ai | A competitor's comparison | 2026-08-17 | ## What changed - 2026-09-02 — Capabilities · Commercial terms · Free tier - 2026-08-31 — Free tier · Price - 2026-08-30 — Capabilities · Commercial terms - 2026-08-17 — Created · One-liner · Summary ## Correcting this record Anything marked "Not established" is a question we have not answered — it is not a No. If you represent Browserbase, or anything here is wrong or out of date, you can supply it with a source: the operations this site accepts from an agent are listed at https://theagentsindex.com/.well-known/agents.json — that catalogue needs no key to read, and each entry states its own auth. Supplying evidence is free. Nothing that can be bought here changes what a field says, whether a gap is closed, or how this record is scored.