Run this model here

No app to install. 491 MB to get started. Your messages stay on this device.

Best at short drafting and quick rewrites — a friendly message, a shorter version of some text. This is a tiny 0.5B model: it will get facts, math, and anything complex wrong. It's here to prove it runs on your machine, not to be your everyday assistant.

Will it run here?

Checking this browser…
not running
  1. Compatibility check
  2. Getting the model ready
  3. Checking the downloaded file
  4. Starting the model on your computer
  5. Ready for your message
  6. Answering on your device
  7. Answered on this device
Prove the safety gate (developer check)

Loads with a deliberately wrong expected hash. The file must be refused before it ever reaches the model — a fail-closed check.

What exactly is running

Identity and hash are resolved live from Hugging Bay's download-plan; the file is verified against that hash in your browser before it runs.

Want more than this small model?

This in-browser run is the smallest, zero-install rung. For bigger models Hugging Bay offers a guided LM Studio path (guidance, not yet a proven run) and a proven terminal command for the exact bytes. Each artifact's download page lists the most capable honest option for your machine.

Your run receipt (personal proof)

This is your own in-browser success — proof that the verified file ran through the local runtime on this device. It is not a claim by Hugging Bay, and it is not a statement about answer quality.

(run something to populate this)