# Browsa > Browsa is a cloud browser API for AI agents. One API call starts a real cloud browser that you drive with Playwright, Puppeteer or raw CDP, or you describe a task in plain language and an agent runs it and returns JSON. It also offers one-call scrape, extract and screenshot, schedules with signed webhooks, persistent identities (saved logins), a live view of every browser, Python and TypeScript SDKs and an MCP server. Pay-as-you-go credits that never expire; 100 free credits after email verification, no card. Key facts: a browser costs 1 credit per started 12 minutes (5 credits for a full hour); a credit is $0.10 on the $20 Starter pack, $0.083 on Growth ($100) and $0.067 on Pro ($500). Sessions last up to 1 hour; a workspace has 6 browser slots. Browsa runs Phantom, its own browser. Proxies: Browsa's network costs 5 credits per GB (datacenter) or 50 credits per GB (residential); your own HTTP, HTTPS or SOCKS5 proxy is free; without one, browsers use Browsa's data-center IP. CAPTCHA solving is not offered. API base URL: https://browsa.io/v1. Authenticate with `X-API-Key: agt_live_...` or `Authorization: Bearer agt_live_...`. SDKs: `pip install browsa`, `npm i browsa`. MCP server: `npx -y @browsa/mcp`. Support: support@browsa.io ## Full summary - [llms-full.txt](https://browsa.io/llms-full.txt): the whole product, API surface, pricing, limits, measured results, comparisons and FAQs in one Markdown file ## Docs - [API documentation](https://browsa.io/docs): browsers over CDP, persistent identities, agent tasks, sessions (follow-up messages on the same browser), quick scrape/extract/screenshot/batch/crawl, jobs (outcome, signals, step timeline), schedules, webhooks, saved proxies, saved LLM keys and credentials, credits, budgets, errors, rate limits - [Models](https://browsa.io/guides/models): the hosted tiers `browsa`, `browsa-fast` and `browsa-smart` (what each runs and its credits per task, live at `GET /v1/models`), your own provider key, OpenAI-compatible servers, model name forms, the `reasoning` field, token usage and estimated cost on a job, and the model checks made before a browser starts - [OpenAPI spec](https://browsa.io/openapi.yaml): OpenAPI 3.1 - [Cookbook](https://browsa.io/cookbook): copy-paste SDK recipes ## Guides - [What is a cloud browser for AI agents?](https://browsa.io/guides/what-is-a-cloud-browser): definition, how it works, cloud vs local headless Chrome, what to check before choosing - [Playwright and Puppeteer over CDP](https://browsa.io/guides/playwright-puppeteer-cdp): connect, reconnect, download files, viewport rules, proxies, saved logins - [MCP server setup](https://browsa.io/guides/mcp-server): Claude Desktop, Claude Code, Cursor, VS Code; tools and costs - [browser-use](https://browsa.io/guides/browser-use) - [Stagehand](https://browsa.io/guides/stagehand) - [OpenAI Agents SDK](https://browsa.io/guides/openai-agents-sdk) - [Claude computer use](https://browsa.io/guides/claude-computer-use) ## Pricing - [Pricing (Markdown)](https://browsa.io/pricing.md): credit packs, price per unit, task formula, refunds, limits, worked cost examples, balance API - [Pricing page](https://browsa.io/pricing) ## Comparisons - [Browserbase alternatives](https://browsa.io/vs/browserbase-alternatives): Browsa, Browserbase, Steel and Browser Use Cloud side by side, with sources - [Browsa vs Browserbase](https://browsa.io/vs/browserbase) - [Browsa vs Steel](https://browsa.io/vs/steel) - [Browsa vs Browser Use Cloud](https://browsa.io/compare) ## Use cases - [All use cases](https://browsa.io/use-cases) - [Run Playwright in the cloud](https://browsa.io/use-cases/cloud-playwright) - [Puppeteer in the cloud](https://browsa.io/use-cases/puppeteer-cloud) - [Browser agents in plain language](https://browsa.io/use-cases/browser-agents) - [A real browser for Claude and Cursor (MCP)](https://browsa.io/use-cases/mcp-browser-claude-cursor) - [Structured data extraction](https://browsa.io/use-cases/web-scraping-structured-data) - [Price and stock monitoring](https://browsa.io/use-cases/price-monitoring) - [Competitor monitoring](https://browsa.io/use-cases/competitor-monitoring) - [Scheduled browser jobs](https://browsa.io/use-cases/scheduled-browser-jobs) ## Optional - [Phantom](https://browsa.io/phantom): Browsa's own browser: what it can do, measured values and known gaps - [Benchmarks](https://browsa.io/benchmarks): dated bot-check and real-site results, misses included - [Status](https://browsa.io/status) - [Terms](https://browsa.io/terms) - [Privacy](https://browsa.io/privacy)