JevMade Sign in
← Back to experiments

Playable demos & bots

AutoTrust JEV-27B-VL

AutoTrust’s open-weight JEV-27B-VL turns images into fixed-answer decisions, with a sped-up Mario demonstration showing movement and jump choices.

ACTUAL RECORDING · 12 SEC

What it does

This is a separate model, not TypeSafe Jev. The linked demo repository does not include the Mario controller, and the maker’s 5× playback is not a real-time speed test. Image confidence has not been systematically measured.

How you can use it

To try visual decisions in your own app, ask a developer to study the image example. Send pictures and a question with a fixed set of answers to your own model server. The example ranks video covers; the linked code does not include the Mario controller.

AutoTrust’s model is separate from TypeSafe Jev. Hosting it requires a graphics card with at least 80 GB of memory. Test its answers on your own images: the maker has not systematically measured image confidence, and a probability is not permission to act.

Keep this for later

Sign in to bookmark experiments, guides and videos, and keep notes only you can see.

Continue to sign in

We’ll bring you back to this listing.