Oh my AI! — Chat with AI Directly in Your Browser
Discover Oh my AI!, our open-source chat application that runs large language models entirely in your browser using WebLLM and WebGPU. No server, no data sharing — just private AI conversations.
Oh my AI! — Chat with AI Directly in Your Browser
What if you could have a conversation with an AI assistant without any of your data ever leaving your computer? That’s exactly what Oh my AI! does.
What is Oh my AI!?
Oh my AI! is an open-source web application that lets you chat with large language models (LLMs) entirely in your browser. No cloud server, no API key, no account required. You open the page, pick a model, and start chatting.
Behind the scenes, the app uses WebLLM — an open-source project by MLC — to run AI models directly on your device using WebGPU, the modern browser API for GPU-accelerated computing.
Why Does It Matter?
Most AI chatbots send your messages to remote servers for processing. That means your conversations — personal questions, sensitive documents, creative ideas — all pass through someone else’s infrastructure.
With Oh my AI!, everything stays on your machine:
- Complete privacy — Your conversations never leave your browser. No server logs, no data collection, no third-party access.
- Works offline — Once the model is downloaded, you can chat without an internet connection.
- No account needed — Just open the page and start talking. No sign-up, no subscription.
- Open source — The entire codebase is available on GitHub. Inspect it, fork it, improve it.
How Does It Work?
The magic comes from two open-source technologies:
- WebLLM (MLC) — A JavaScript library that loads and runs optimized LLM models in the browser. MLC compiles models into a format that can run efficiently on consumer hardware.
- WebGPU — A modern browser API that gives web applications direct access to your GPU. This is what makes running billion-parameter models in a browser tab actually feasible.
When you first use Oh my AI!, the app downloads a model to your browser’s cache. After that, inference happens locally on your GPU — no round-trip to a server.
What Models Can I Use?
Oh my AI! supports several open-source models optimized for in-browser execution, including various sizes of Llama, Mistral, Gemma and Phi. You can choose the model that best fits your hardware and needs.
Try It Now
Head over to ohmyai.org and start chatting. All you need is a modern browser with WebGPU support (Chrome, Edge, or any Chromium-based browser).
And if you’re curious about the code, everything is on GitHub. Contributions are welcome!