AI Chess Leaderboard (Reasoning) — leaderboard
LLMs play full chess games from the starting position. Ranked by Elo rating derived from head-to-head matches. Tests legal move generation, multi-step planning, and game-state tracking over 264 evaluated models.
Metric: Elo. Source: dubesor.de. Status: saturated. 300 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 1922 |
| 2 | Gemini 3 Pro (Preview) | 1850 |
| 3 | Qwen 3 Max (Thinking) | 1800 |
| 4 | GPT-5 Codex | 1777 |
| 5 | Gemini 3.5 Flash | 1772 |
| 6 | Gemini 3.6 Flash | 1763 |
| 7 | GPT-5.1 Codex | 1743 |
| 8 | GPT-5.5 | 1720 |
| 9 | Claude Fable 5 | 1670 |
| 10 | GPT-5.6 Sol | 1604 |
| 11 | Grok 4 | 1596 |
| 12 | O3 | 1585 |
| 13 | GPT-5 | 1572 |
| 14 | Hy3 | 1532 |
| 15 | GPT-5.1 | 1527 |
Interactive version: theaggregate.ai/benchmark?slug=ai-chess-leaderboard-reasoning · How the rankings work · Data refreshed daily, snapshot 2026-07-22.