Web Llm Chat

WebLLM Chat is a browser based chat application that runs large language models directly on the user's device. The models execute natively in the browser rather than on a remote server, so conversations are private and need no server connection after the model loads. The default model shown on the home page is Llama 3.2 1B Instruct in a quantized MLC build. The interface is a conventional chat workspace. Users can start new conversations, edit a conversation, set a system prompt, clear context, and keep a library of prompts with quick prompt shortcuts. A settings panel and a model selector let the user choose which locally run model answers. The tool suits people who want to try open models without installing software, who care that prompts stay on their own machine, or who want to see what in browser inference can do on their hardware. Performance depends on the device and browser, since the model weights are downloaded and run locally. The app is free to use, and the site shows no paid plans or account requirement.