Local AI Lab

Runs locally

A browser-local AI capability and experiment status entry.

Experimental, coming soon

Run small models in the browser

This MVP shows the model catalog and device state. It does not download weights or run inference. Saving a downloaded model will require separate consent.

Device check

Checking

SmolLM2 360M

约 270 MB
Short text rewriting

Qwen2.5 0.5B

约 400 MB
Chinese and English text tasks

Model weights will be stored persistently only after explicit permission.

Source & licensesopen-source components and licenses

Experimental · Data stays in your browser

About this tool

How to use

  1. Open the page — device capabilities (WebGPU, etc.) are detected automatically.
  2. Check experiment status and what is planned next.

Features

  • Local device AI capability detection
  • Experiment status at a glance
  • Experimental, with honest capability boundaries

Privacy

Capability detection runs locally; no device information is uploaded.

FAQ

Can it run models today?

Not yet. Real model download and inference are on the roadmap and will be announced separately.