JevMade Sign in
← Back to experiments

Playable demos & bots

Human vs JEV · Table Tennis

A table-tennis game where Jev's answer is the paddle's target pixel and a tiny local servo simply moves there.

Source screenshot of Human vs JEV · Table Tennis
SOURCE SCREENSHOTFull screenshot ↗

What it does

Pure mode has no trajectory predictor. The maker reports an 88% hit rate after adding time-to-arrival context.

How you can use it

Sketch your game court by dividing the paddle movement area into numbered landing zones. Your developer can write a script that sends the ball speed, position, and travel time to the model. The model then chooses the best zone, and basic game code slides the paddle there.

Maker-reported (not independently measured by JevMade): 88% hit rate at a 1.9 s median latency (author-reported) · About 700 input tokens per call at $0.042/M tokens, roughly $0.00003 per decision (author-reported) · Adding a time-to-arrival sentence to the state raised the hit rate from 50% to 88% (author-reported)

Primitives
choice
Platform
Python
Added
Project created
GitHub stars
1 (snapshot, not live)