JevPlay
Sign In
All Jev apps and community cases

Jev community / Jev in games

Jev in games

Game integrations make the decision loop visible: a board or world changes, the application supplies the available moves, and Jev selects an action. These six community posts show different ways to prepare that state, from structured boards to game-specific emulator adapters. They are examples of integration design, rather than proof that one model wins every game.

New to the model? What is Jev AI explains its state-to-decision interface.

Browse Jev case topics

From game state to a legal move

  1. Observe the board, player position, resources, or other game variables that matter for the next move.
  2. Give Jev a bounded action space, such as play or discard, a direction, or a set of legal placements.
  3. Let game code validate and execute the selected action, then observe the updated state before the next decision.

What to compare across game demos

Look at the observation adapter, legal-move filtering, decision frequency, and scoring rules before comparing results. A precise RAM adapter and a general visual agent solve different problems. Latency, cost, scores, and outcomes in the original posts are their authors’ reports; JevPlay has not independently reproduced these community benchmarks.

Selected community demos

Each selection links to its original author’s post. Source text and media come from the existing Jev community archive.

Balatro: play, discard, and shop

Juris describes an adapter that supplies cards, jokers, and the blind before asking Jev to choose play or discard. The same run also includes shop decisions after the ante.

@Juris_Savos · 2026-09-18 · Original post
001en
Juris@Juris_Savos

Jev from @typesafeai plays @BalatroGame. Every hand it sees the cards, jokers, and blind, then picks play or discard in ~200–500ms. After the ante, it shops too. A full run costs less than a cent. https://t.co/PyGvC0dvga

1.5Kviews4likes2saves
Translation

An arcade experiment with manual and AI control

Adrian’s demo switches between manual and AI-driven play. The post focuses on expressing ball protection, powerups, and risk as variables and action criteria that the decision model can use.

@adrianmg · 2026-09-18 · Original post
002en
Adrián Mato 🐙@adrianmg

Had some fun taking Jev from @typesafeai for a spin, switching seamlessly between manual and AI-driven gameplay. It’s clocking 2–3 calls/sec at 60 FPS, with ~150ms avg latency. The most interesting part wasn’t the speed, though. It was distilling the game into...

646views14likes4saves
Translation

Snake: two Jev-controlled players

Raihan Khan shares a Snake project in which two Jev instances compete. The original post includes a playable link and describes the project as open source.

@raihankhan_rk · 2026-09-18 · Original post
003en
Raihan Khan@raihankhan_rk

I've had access to Jev by @typesafeai for about 30 hours now and I can't stop thinking of the million places it can be used 🤯 Meanwhile, checkout this fun snake game I built where 2 Jev's compete with each other to take the trophy 🏆 Try it out here 🔗 https:...

3.3Kviews16likes13saves
QuoteLink 2Translation

Tetris with shared pieces and legal moves

Ashutosh compares Jev with two other models using the same piece sequence and legal moves. This is a useful example of specifying the comparison setup before interpreting a reported score.

@ashutoshftw · 2026-09-17 · Original post
004en
Ashutosh Mathore@ashutoshftw

Had jev, claude haiku 4.5, and gemini 3.5 flash-lite play tetris. Same 200 pieces, same legal moves, real time. Jev: 9200 pts, 75 lines, 300ms/move, 0 errors Claude: 8900 pts, 72 lines, 1.52s/move, 0 errors, $0.48 Gemini: 9000 pts, 75 lines, 1.13s/move, 0 erro...

20.2Kviews122likes41saves
Translation

Mario: structured RAM and emulator lookahead

TheINAOG parses emulator RAM, simulates controller macros, and filters predicted deaths before asking Jev to choose. The author explicitly notes that this game-specific adapter does not solve general visual perception.

@TheINAOG · 2026-09-17 · Original post
006en
@TheINAOG@TheINAOG

Finally I got access to @typesafeai (thank you guys!) Jev + emulator lookahead cleared Super Mario Bros. 1-1 only 0.04$. In this setup, Jev sees no pixels: I parsed RAM into structured state, simulated controller macros, filtered predicted deaths, and let Jev...

296views2likes1saves
Translation
Play the JevPlay games

Try daily challenges with shared rules, a visible decision trace, and a human-versus-Jev comparison.

JevPlay is an independent project and is not affiliated with TypeSafe. Community posts, media, and trademarks belong to their original owners. Browse the full Jev case library for more sources.