| Name | Modified | Size | Downloads / Week |
|---|---|---|---|
| Parent folder | |||
| chatllm-v3-dark.jpg | 2026-08-11 | 46.4 kB | |
| chatllm-v3-home.jpg | 2026-08-11 | 50.9 kB | |
| chatllm-v3-mobile.jpg | 2026-08-11 | 27.5 kB | |
| ChatLLM Web v3.0.0 source code.tar.gz | 2026-08-11 | 944.1 kB | |
| ChatLLM Web v3.0.0 source code.zip | 2026-08-11 | 967.2 kB | |
| README.md | 2026-08-11 | 1.2 kB | |
| Totals: 6 Items | 2.0 MB | 0 | |
ChatLLM Web v3.0.0
ChatLLM is now a complete browser-native Local Model Studio built around an agent-ready Beautiful UI component system.
Highlights
- Five curated WebLLM models with device-aware recommendation and 4K contexts.
- New
/modelsstudio for WebGPU insights, compatibility, caching, loading, switching, deletion, and custom MLC manifests. - Beautiful UI–based Chat Workspace, Prompt Bar, activity trace, recommendations, approvals, context cards, insight cards, and model catalog.
- Local TXT, Markdown, JSON, and code context with explicit limits.
- IndexedDB conversations, messages, attachments, preferences, and custom models.
- English and Simplified Chinese, light and dark themes, responsive mobile layouts, and reduced motion.
- Glimm WebGL sweep accents and Motion transitions.
- The original handwritten v1 inference core preserved under
legacy/early-webllm-runtime.
Model catalog
- Llama 3.2 1B
- Qwen 3.5 0.8B
- Qwen 3.5 2B
- Qwen 3.5 4B
- Qwen 2.5 Coder 3B
Local data
v3 starts with a fresh chatllm-v3 IndexedDB schema. The v2 localStorage conversation store remains separate.
Links
- Live: https://chatllm-web.pages.dev
- Documentation: https://github.com/Ryan-yang125/ChatLLM-Web#readme