Who should use this model?
- Trying a small local model in a supported browser.
- Short, simple writing and question-answering experiments.
Choose something else when
- You need dependable factual, legal, medical, or complex coding answers.
- Your browser or graphics hardware does not support WebGPU.
What hardware makes sense?
The download is only part of the memory needed. Leave room for the app, the input it processes, and other open programs. For chat models, context includes the instructions, conversation, and answer. Longer context needs more memory.
- CPU only
- Runtime support must be checked for the selected artifact.
- 8GB system memory
- Check the selected runtime and conversation length.
- 16GB system memory
- Check the selected runtime and conversation length.
- 32GB system memory
- Check the selected runtime and conversation length.
- 8GB GPU memory
- The small webllm build still requires a compatible webgpu browser.
- 12GB GPU memory
- Check webgpu compatibility.
- 24GB GPU memory
- Check webgpu compatibility.
- Apple unified memory
- Check webgpu compatibility and available unified memory.
Model facts that change the decision
- Parameters
- 0.6B
- Input
- Text
- Published context limit
- 32,768 tokens
- Model format
- Safetensors · MLC
- Suggested compression
- The browser demo uses the WebLLM q4f16_1 or q4f32_1 build.
- Languages
- English · Multilingual
Try the browser model
The browser demo uses its own packaged model and context setting. Check browser support and the download before starting; the model card’s maximum context does not establish the demo’s limit.
Conditions that change the answer
- This small model can make mistakes and has less capacity than larger models.
- The browser demo uses a 4,096-token context; the upstream model card describes a 32,768-token context.
- Browser-model downloads and GPU memory allocation are separate from the size of a Safetensors model.
Read the controlling license text →