Have a ChatGPT Subscription and a Windows PC? Now Give It a Browser.

ChatGPT Pro gives you GPT-5 Codex with a 400K context window. GitHub Copilot gives you Claude Sonnet 4.6 and Gemini 3 Pro. Qwen Code gives you a million-token context window.

These are some of the most capable AI models available today. And starting with BrowserOS v0.44.0, they can all operate your browser.

Sign in with your existing account. Pick a model. Tell BrowserOS what you need done. Your coding agent handles the rest — navigating pages, clicking buttons, filling forms, extracting data, and working across multiple tabs.

No API keys. No extra cost. Just your subscription, now with hands in the browser.

Connect in 30 Seconds

We built this around OAuth — the same sign-in flow you use everywhere else.

ChatGPT Pro / Plus — Sign in with your OpenAI account. Access GPT-5.4, GPT-5.3 Codex, and the full Codex lineup with up to 400K context. The same models powering ChatGPT's most advanced features, now automating your browser.

GitHub Copilot — Sign in with your GitHub account. Free tier users get GPT-5 Mini, Claude Haiku, and GPT-4o. Pro subscribers unlock Claude Sonnet 4.6, Claude Opus 4.6, Gemini 3 Pro, GPT-5.4, and 15+ more models. One GitHub subscription, access to models from OpenAI, Anthropic, and Google — all working in your browser.

Qwen Code — Sign in with your Alibaba Cloud account. Access Qwen 3 Coder Plus and Qwen 3.5 Plus with a 1 million token context window — the largest of any provider we support. Ideal for processing long documents, entire documentation sites, or working across dozens of tabs without losing context.

Open Settings > AI Providers, click Connect, sign in, and you're ready.

What BrowserOS-Level Productivity Looks Like

These models are powerful on their own. Paired with BrowserOS — which gives them native access to the DOM, accessibility tree, screenshots, keyboard, and mouse — they become something more.

Research at scale. Open 15 competitor pages, extract pricing and feature data from each, and get a structured comparison — without copying and pasting a single thing.

Forms that fill themselves. Point the agent at a multi-step application form, give it your information once, and watch it navigate dropdowns, checkboxes, and conditional fields across pages.

Web workflows on autopilot. "Every morning, check these 5 dashboards, screenshot the key metrics, and compile a summary." Set it up once. The agent runs it on schedule.

Multi-tab coordination. The agent doesn't just work in one tab. It opens new tabs, switches between them, groups related tabs, and maintains context across all of them. Research, compare, and act — the way you would, but faster.

13 Providers, One Browser

These three OAuth integrations join a lineup that now covers every way you might want to bring AI to the browser.

Sign in with your subscription — ChatGPT Pro, GitHub Copilot, Qwen Code
Use your own API key — OpenAI, Anthropic (Claude), Google Gemini, Moonshot (Kimi), OpenRouter
Enterprise — Azure OpenAI, AWS Bedrock
Run locally — Ollama, LM Studio
Custom endpoint — Any OpenAI-compatible API

Prefer Claude for reasoning tasks? Use it. Need GPT-5 Codex for complex multi-step workflows? Bring it. Want everything running locally on your machine with Ollama? Go for it.

BrowserOS is open source. You choose the model. We give it the browser.

Privacy First, Always

Your OAuth tokens live on your machine. Requests go directly from BrowserOS to the AI provider — we never proxy, store, or log your conversations. Disconnect a provider and the tokens are immediately deleted.

This is what open source makes possible. You can verify every line of code that touches your credentials. Over 10,000 developers already have.

Get Started

Update to BrowserOS v0.44.0 and connect your first provider: