Run AI Chat Locally in Your Browser
WebGPU brings open models to your own device: no installation, no account - prompts and history stay local.
Local AI without the setup
Running AI models on your own machine used to mean installing tools like Ollama or LM Studio, managing model files, and living in a terminal. Modern browsers changed that: with WebGPU, the browser itself can run open-weight language models on your graphics hardware. Open a page, pick a model, chat - that is the whole setup.
ASD123.ai Local AI Chat uses exactly this approach. The model weights are downloaded once and cached; after that, every conversation runs on your device. No account, no API key, no server that sees your prompts.
What you can do with it
- Choose between open models (including Gemma and Qwen variants) with size and hardware notes per model
- Attach documents and images and ask questions about them - processed in the browser
- Edit earlier prompts, copy or regenerate answers, and export whole chats as Markdown
- Keep chat history in your browser’s local database (IndexedDB) - rename, delete, nothing syncs anywhere
- Stop generation at any time, and manage the context window for long conversations
What you need
A browser with WebGPU (current Chrome or Edge; Safari and Firefox are catching up) and enough memory for the model you pick. Small models run on most modern laptops; larger ones benefit from a dedicated GPU or Apple Silicon. The first load downloads the model weights - from a few hundred megabytes to a few gigabytes - which your browser caches for next time.
Frequently asked questions
Is it free?
Yes. The models are open-weight, the tool is free, and there is no account or API key. Your only cost is the one-time model download and your own hardware’s electricity.
What hardware do I need?
A WebGPU-capable browser (current Chrome or Edge) and enough RAM/VRAM for the chosen model. Small models run on most modern laptops; bigger models are noticeably faster on a dedicated GPU or Apple Silicon.
Where is my chat history stored?
In IndexedDB, a local database inside your browser profile. It never leaves your device, and you can rename or delete chats at any time.
Is it as good as ChatGPT?
Open models that fit on consumer hardware are behind the largest cloud models in capability - that is the honest trade-off. For summarizing, drafting, translating, and working with your own documents, they are more than good enough, and everything stays private.
Try a local model in your browser
No install, no account - just pick a model and start chatting.
Related: Count LLM tokens offline · Anonymize text before AI