Jev Chess Lab
Six recorded chess experiments with Jev via OpenRouter, candidly showing that unaided Jev still blunders pieces.
Overview
Jev Chess Lab records six chess experiments with Jev via OpenRouter, with Jev playing White: independent play with the original prompt, a tactical filter in code, Stockfish assistance, and returns to Jev-only play with longer instructions and history, explicit piece descriptions, and a propose-review-choose loop. Games are rendered as looping GIFs and interactive replays, with research and usage notes in Russian.
The headline is candid: the Stockfish-assisted system won, independent Jev still blunders pieces, and assisted results are not evidence that Jev became a strong player. In the newest experiment Jev proposed, reviewed three moves and chose, changing its first choice on 16 of 23 turns, yet still lost by checkmate after 115 API calls costing about $0.009. The explicit-pieces experiment hit its 60-call budget unfinished, with Jev repeatedly shuffling a rook between d1 and e1. It is GPL-3.0-or-later with third-party notices.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | Which move should White play? | — |
| Choice | After reviewing three candidates, which move is best? | — |
How it works
- Describe the position to Jev in each experiment's format
- Jev chooses a move (optionally filtered or assisted)
- Record games and costs
- Publish replays and findings
Strengths
- Honest negative results
- Clear separation of assisted and unaided play
- Low cost per game
Caveats
- Small number of games
- Chess is outside Jev's intended use
Getting started
Watch the games at https://denikuchero.github.io/jev-chess-lab/.At a glance
- Type
- Project
- Status
- research
- Language
- Python
- Runtime
- experiment scripts and replay site
- Has code
- yes
- License
- GPL-3.0-or-later
Use cases and integrations
How curated lists describe it
Recorded chess experiments with a candid result: Jev on its own still blunders pieces.
MrJev/awesome-jev
Frequently asked questions
Can Jev play chess well?
Not on its own; the experiments show it still blunders pieces.
Did any setup win?
The Stockfish-assisted one, which says nothing about Jev's own strength.
What did the review loop do?
Jev changed its first choice on 16 of 23 turns but still lost.
Related projects
Fifteen live arcade games where Jev and LLMs play each other, led by WikiRace in which Jev picks only real links.
ProjectbetaTypeScriptOne-page web app where Jev plays chess against OpenRouter LLMs, Stockfish or you, showing the moves it weighed and their probabilities.
AppbetaJavaScriptPac-Man where Jev or other System One models choose each move at junctions, with live odds, fallbacks and a leaderboard.
ProjectbetaTypeScriptBrowser chess lab where Jev, LLMs and people play, each Jev turn being one Choice over the legal moves, with Stockfish analysis.
ProjectbetaPythonNiceGUI desktop chess app for evaluating Jev's move choices against Stockfish, random players or itself, with PGN export.
ProjectprototypePythonBrowser 2D driving sim where Jev answers lane, speed, hazard and pedestrian-yield questions every 200 ms, with an LLM comparison mode.
ProjectprototypeJavaScript