Jev-Style Models Control Doom

Nishaanth Reddy gave four Jev-style decision models controls to Doom: Jev, Laya, finetuned ModernCE, and Qwen3.5. A deterministic Python adapter turns ViZDoom’s visible-object labels and HUD into text, and each model chooses whether to turn, move, or fire.

View on X

Nishaanth Reddy@reddmachine𝕏

I gave four Jev-style decision models the controls to Doom: Jev, Laya, finetuned ModernCE and Qwen3.5. A deterministic Python adapter turns ViZDoom's visible-object labels + HUD into text. Each model picks turn, move or fire. No extra Doom-specific training. Same starting seed, separately recorded games. Each model's actions change its view. The video shows action probabilities, kills and survival

Sep 19, 2026X postsView on X

The models play from the same starting seed in separately recorded games. Each model’s actions change its view; the video shows action probabilities, kills, and survival.

Also filed under Games & real time

Benchmarks Jev on Pokémon Red
The project benchmarks Jev, TypeSafe’s fast decision model, against Jev paired with a GPT-6 Sol planner for playing Pokémon Red.
Plain-English Cellular Ecosystem Simulator
LifePot is a cellular automaton-inspired ecosystem that users describe in plain English. Jev turns those inputs into species, feeding relationships, reproduction strategies, and environmental conditions.
Mapping Children's Behavioral Intent in Games
Mores Research used Jev to map children’s behavioral intent in games and examine how it correlates with real-world behavior. It reports mapping intents across 45+ real sessions involving 20 child users over a long horizon.
StarCraft Agent for Fast Tactical Decisions
Jev_Star is a StarCraft agent experiment that uses Jev for fast tactical decisions.