ChessBench: leaderboard
Chess benchmark where language models play rated games against fixed reference opponents, scored on rule-compliant play, move quality, and Monte Carlo Elo.
Metric: Overall Score (%). Source: chessbench.ai. Status: years away from saturation. 61 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.8 Flash | 75.9 |
| 2 | GPT-5.5 | 62.9 |
| 3 | Gemini 3.7 Flash | 62.8 |
| 4 | Gemini 3.1 Pro (Preview) | 56.4 |
| 5 | GPT-5.6 Sol | 54.8 |
| 6 | Gemini 3.6 Flash | 54.8 |
| 7 | Gemini 3.5 Flash | 52 |
| 8 | Muse Spark 1.3 | 51.8 |
| 9 | GPT-5.4 | 47.7 |
| 10 | Muse Spark 1.2 | 42.7 |
| 11 | Claude Fable 5 | 41 |
| 12 | O3 | 39.2 |
| 13 | GPT-5.1 | 37 |
| 14 | Muse Spark 1.1 | 36.8 |
| 15 | GPT-5.6 Terra | 36 |
Interactive version: theaggregate.ai/benchmark?slug=chessbench · How It Works · Data refreshed daily, snapshot 2026-09-19.