What it does
Trained on 2.2M frames under a Jev System One-shaped contract, the confidence gates an optional handover to a search program, and the weights plus a ten-game demo are public.
Benchmarks & research
PlayJev
A 0.8B vision-language model playing ten browser games straight from pixels: one frame in, one move out, 43 ms on an H200.
Bookmark: PlayJev
Keep this in your collection.
Leave a noteWhat would you try with this?
: PlayJev
Only you can see your notes.
Screenshot unavailable. Open the experiment ↗
How you can use it
You can use this to build an AI that plays simple visual games. Your developer gives the model an image and a list of possible moves. The model looks at the screen and picks a move. This skips the need to read complicated game code.
You can also borrow its approach to asking for help. The software checks how certain the model is about a move. If the AI is unsure, it hands the choice to a separate backup program. Your developer needs a capable graphics card to run this model.