Jev Chess Decision Model

Jev played chess against Erwin’s local Qwen 3.8 27B stack with reasoning off, using one parallel pass per decision and no reasoning. The game ended in a draw after ten checks, with Jev holding a rook, bishop, and three pawns against a lone king.

View on X time337 ms/move

Erwin@eres2k𝕏

Jev isn't an LLM — one parallel pass per decision, no reasoning at all. I put it on a chess board vs my local Qwen 3.8 27B (reasoning:off) stack on my 5090. Playing on pure single-pass judgment, Jev navigated a tactical bloodbath and came out with rook + bishop + 3 pawns vs a lone king. +11 material against a full language model. Then ten checks, no mate. Draw. 337 ms/move, twice as fast my local

Sep 17, 2026X postsView on X

The match ran on Erwin’s 5090. Jev had an 11-point material advantage but did not find mate.

Also filed under Games & real time

Steer a 3D Box with Goals
The build steers a 3D box with plain-language goals. An LLM plans each instruction, while TypeSafe Jev makes move, turn, and jump decisions in 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.
Jev-Powered Rust Survival Game
Anthony Eckert says he has a Rust survival-game clone in pre-alpha, with an agent playing using Jev. He says crypto rewards are coming soon, with browser play and full agent support planned for tomorrow.
A Jev-Made Unity Build
Kevin Kern says the build was made with Jev, GPT Live, Blender, Unity, and more.