Browserbase
Also known as: Stagehand, Browserbase MCP
Browser infrastructure platform for AI agents, providing hosted browsers, the Stagehand SDK, and Browserbase MCP integration.
Browserbase is cloud infrastructure that gives AI agents a real web browser to work in. Its founding premise is simple: if you want an AI to do the same things you do on the web, it needs a browser, because much of the web, the parts behind logins, heavy JavaScript, CAPTCHAs, and interactive flows, is not reachable through clean APIs. Running and scaling a fleet of headless browsers reliably is hard, so Browserbase operates that fleet as a service and lets developers spin up managed Chromium sessions through a single API rather than maintaining their own browser servers.
The core offering is large-scale headless browser infrastructure: globally distributed, isolated sessions that start in milliseconds and scale to many concurrent browsers, with the operational details handled for you. It works as a drop-in for the standard automation frameworks, Playwright, Puppeteer, and Selenium, so teams can move existing scripts to the cloud with minimal changes, and it adds the things that break agents in the wild: stealth fingerprints and proxy support to get past bot detection, plus rich observability through session recordings, a live view, and an inspector for debugging flaky flows and login walls.
On top of the raw browsers, Browserbase builds an agent platform with several primitives under one API key. Agent Identity helps agents get past auth walls and bot detection through partnerships, Search and Fetch APIs give agents fast, token-efficient web context, Functions run agent code right next to the browser for low latency, and a Model Gateway provides access to major models through a single key and bill.
The company also maintains Stagehand, an open-source SDK that is the intelligent layer many teams pair with the infrastructure. Instead of brittle selectors that break when a site changes its markup, Stagehand lets you write browser actions as plain-English instructions that an AI resolves at runtime and that self-heal across page redesigns, with primitives to act, extract structured data, and observe a page. Importantly, Browserbase is infrastructure rather than an agent itself: it supplies the browsers, identity, and observability, while the agent logic comes from Stagehand, frameworks like Browser Use, or a team's own code.
Vendor details
Canonical URL
https://www.browserbase.com/
Category
Browser / computer-use agent
Company status
independent
Use cases & customers
Target customers
Deployment options
In practice
Your agent needs to operate real websites at scale, but maintaining a fleet of headless browsers with proxies and anti-bot handling is eating your team. Browserbase runs managed Chromium sessions through one API so you don't have to.
Your Playwright scripts keep breaking when sites change their markup. Browserbase's open-source Stagehand SDK lets you write actions as plain-English instructions that an AI resolves at runtime and self-heal across page redesigns.
Your agent keeps hitting CAPTCHAs and login walls, and you can't see why runs fail. Browserbase adds stealth fingerprints, Agent Identity to get past bot detection, and session recordings and a live view for debugging.
Agentic Index coverage score
12.0 / 14 capabilities · 86%
| Integrations & Tool CallingAgentic AI Index Expansion Evidence Report | Full |
|---|---|
| Workflow OrchestrationAgentic AI Index Expansion Evidence Report | Full |
| Knowledge Grounding & RAGAgentic AI Index Expansion Evidence Report | Unable to verify |
| Human Oversight & GuardrailsAgentic AI Index Expansion Evidence Report | Full |
| Security, Identity & GovernanceAgentic AI Index Expansion Evidence Report | Full |
| Observability & AuditabilityAgentic AI Index Expansion Evidence Report | Full |
| Memory & State PersistenceAgentic AI Index Expansion Evidence Report | Partial |
| Deployment & Data ResidencyAgentic AI Index Expansion Evidence Report | Full |
| Prebuilt Agents, Templates & PacksAgentic AI Index Expansion Evidence Report | Full |
| Triggers & Channel CoverageAgentic AI Index Expansion Evidence Report | Partial |
| Model Flexibility & RoutingAgentic AI Index Expansion Evidence Report | Full |
| APIs, SDKs & MCP ExtensibilityAgentic AI Index Expansion Evidence Report | Full |
| Testing, Debugging & OptimizationAgentic AI Index Expansion Evidence Report | Full |
| Browser & Computer UseAgentic AI Index Expansion Evidence Report | Full |
The Agentic Index coverage score grades every vendor Full, Partial or Unable to verify against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded
Recent platform changes
Browserbase launched Stagehand v4, migrating the framework's core state management, target tracking, and CDP dispatch into a browser extension. This architectural change allows remote browser sessions to execute commands natively on the page rather than relying on a lagging client-side mirror. The release also achieves full feature parity across TypeScript, Python, and a newly introduced Go SDK.
Bears on: MCP / tool calling / API
View sourceBrowserbase launched Browserbase Agents: fully managed web agents invoked with a single API call. Instead of provisioning browser infrastructure and wiring an agent loop, developers describe the task and Browserbase runs the browsing, navigation, and extraction end to end on its hosted stack.
Bears on: Browser/computer use
View sourceStagehand server-v3 v3.7.2 added Azure Entra model authentication support.
Bears on: Security / enterprise
View sourcePricing
From $20/mo · free tier
hybrid
Included quota
Free ($0): 1 browser hour, 1-3 concurrent browsers, 15-min max session, 7-day retention, 1 project, free Functions, email support. Developer ($20/mo): 100 browser hours, 25 concurrent, 1GB proxy, 6-hr sessions, Basic Stealth + auto CAPTCHA, 30-day retention, 2 projects. Startup ($99/mo): 500 browser hours, 100 concurrent, 5GB proxy, 6-hr sessions, Basic Stealth + auto CAPTCHA, 30-day retention, 5 projects, priority support. Scale (custom): 250+ concurrent, flexible usage, 6+ hr sessions, Advanced Stealth, 30-90+ day retention, HIPAA/SOC 2/SSO, Slack/live-chat support.
What is public
Browserbase (browserbase.com - cloud headless-browser infrastructure for AI agents + web automation; creators of the Stagehand AI browser-SDK) uses HYBRID pricing: subscription tiers + usage-based browser hours/proxy. Free: $0/mo (1 browser hour, 1-3 concurrent, 15-min sessions, 7-day retention). Developer: $20/mo (100 browser hours, 25 concurrent, 1GB proxy). Startup: $99/mo (500 browser hours, 100 concurrent, 5GB proxy, priority support). Scale: custom (250+ concurrent, Advanced Stealth, HIPAA/SOC 2/SSO). Sessions bill by the minute; Functions (code execution) are free on all plans; BYO proxies supported.
Billing mechanics
Each paid tier bundles INCLUDED browser hours + proxy bandwidth (renewing monthly); beyond the bundle you're not cut off - you pay usage OVERAGE: browser hours at $0.12/hr (Developer) or $0.10/hr (Startup), proxy at $12/GB (Developer) or $10/GB (Startup), plus per-GB-second compute ($0.0000167 Developer / $0.00001389 Startup). Sessions are billed by the minute (first minute rounded up). Concurrency, session-length, retention, and stealth level scale with the tier. Scale is fully custom/usage-based.
Cost watchouts
Three metered dimensions stack - browser hours + proxy GB + GB-second compute - so long or proxy-heavy runs add up beyond the base fee; cold starts take 5-10s per session (overhead on short jobs); session-minute rounding (first minute up) penalizes very short tasks; Advanced Stealth + compliance (HIPAA/SOC 2/SSO) are Scale-only; concurrency is plan-capped (1-3 free to 25 Developer to 100 Startup)
Variable cost rationale
Hybrid and genuinely usage-driven - a flat tier fee plus metered browser hours, proxy GB, and compute GB-seconds; cost scales with session length, concurrency, and proxy use, so heavy/long automation runs can far exceed the base subscription
Additional watchouts
Usage metering across hours/proxy/compute makes long or heavy runs costly (model it before scaling); cold-start latency (5-10s) hurts short jobs; cloud-only (no true on-prem); Advanced Stealth + compliance gated to Scale; free-tier limits (1 browser hour, 15-min sessions) are for testing only; vague agent prompts cause missed clicks (a Stagehand/agent-quality caveat, not pricing)
Overage / add-ons
Developer/Startup don't cut you off at the included limit - browser hours continue at the plan's overage rate ($0.12/hr Developer, $0.10/hr Startup), proxy at $12/$10 per GB, plus GB-second compute charges; included hours + proxy renew monthly. Scale negotiates flexible usage-based rates.
Sales call required
Mixed (some tiers require a call)
Free / trial
Free (no card required)
Lowest paid plan
n/p (free tier; paid floor not captured)
Commercial notes
SF startup founded by Paul Klein; AI-agent-first positioning (the 'eyes and ears' for an LLM 'operating system'); creators/maintainers of Stagehand (the leading AI browser-automation SDK, ~20k stars - Playwright control + act/extract/observe primitives); funded across multiple rounds (Kleiner Perkins-led, ~$67.5M total reported); competes with Steel.dev (most generous free tier - 100 hrs), Browserless (self-host-friendly), Scrapfly (broad framework support), Bright Data/Oxylabs (proxy-scale enterprise), Browser Use
Key ambiguities
Pricing was restructured - the older tier set ('Hobby $39/mo, 200 browser hours') was replaced by the current Free/$20 Developer/$99 Startup/Scale lineup (verified June 2026); free-tier concurrency is cited as 1 (some sources) or 3 (others); session billing granularity is described as per-minute (docs) or per-30-seconds (some reviews); Scale/Enterprise pricing is custom
Cancellation / refund
Free tier (no card) for testing/one-offs; Developer/Startup self-serve monthly with usage overage (no hard cutoff); included hours/proxy renew monthly (no rollover implied); Scale is custom-contracted (compliance, dedicated support); BYO proxies reduce proxy spend
Support SLA / resale
Email support (Free); standard support (Developer); priority support (Startup); Scale adds high-priority support (Slack/live chat), HIPAA/SOC 2/SSO compliance, dedicated proxies, and 90+ day retention; cloud-only (no on-prem, though consulting for hybrid architectures is offered); global multi-region infrastructure; integrates with Playwright/Puppeteer/Selenium + Stagehand SDK (LangChain/CrewAI/Mastra)
Missing data
Scale pricing is custom. Free-tier concurrency and session-billing granularity are cited inconsistently across sources. The published paid floor: Developer $20/mo (100 browser hours, then $0.12/hr), Startup $99/mo (500 hours, then $0.10/hr), Scale custom - a hybrid subscription + usage model (browser hours + proxy GB + compute). The older 'Hobby $39' tier was replaced.
Related vendors
- AgentQL — A self-healing query language for the web: natural language…
- Airtop — Cloud browsers for AI agents that compile plain English workflows…
- Asteroid — Supervised browser and computer use agents for regulated back…
- Astral — Browser based AI marketing agent that operates social platforms like…
- Automat AI — Managed AI agents that operate computers visually to replace legacy…
- Autotab — General AI agent that uses a mouse and keyboard like a person to do…
Alternatives to Browserbase
The closest documented capability profiles to Browserbase among browser and computer-use agents tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.
- Browser Use11.0 / 14A lighter documented profile than BrowserbaseBrowserbase vs Browser Use →
- Emergence AI12.5 / 14Adds documented Knowledge Grounding & RAGBrowserbase vs Emergence AI →
- Stagehand10.5 / 14Adds documented Knowledge Grounding & RAG
- Island11.0 / 14Adds documented Knowledge Grounding & RAGBrowserbase vs Island →
- Scrapybara9.0 / 14A lighter documented profile than BrowserbaseBrowserbase vs Scrapybara →
- Asteroid9.5 / 14Fuller documented coverage on Triggers & Channel Coverage
Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded