# ChessBench LLM chess results

Ranks LLMs by their ability to plan, predict and adapt.

Source: https://chessbench.org/data
Updated results: 2026-10-26T23:44:16.023Z
Retrieved: 2026-10-28T01:30:18.577Z
Protocol: chessbench-v3

ChessBench Elo ratings compare LLMs within this benchmark only. They are not equivalent to Chess.com, FIDE, or other chess ratings. Beating a 2000-rated LLM here does not establish your rating or predict the same result against a 2000-rated human player.

| Model and thinking | Rank | Elo | Rated games | Wins | Draws | Losses | USD/game | Seconds/turn |
| --- | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: |
| [Gemini 4 Argon (Medium)](https://chessbench.org/models/gemini-4-argon) | 1 | 2170 | 70 | 65 | 1 | 4 | 0.4066 | 9.3 |
| [Gemini 4 Argon (High)](https://chessbench.org/models/gemini-4-argon) | 4 | 2088 | 70 | 61 | 2 | 7 | 0.7215 | 15.0 |
| [Claude Opus 5.5 (xHigh)](https://chessbench.org/models/claude-opus-5.5) | 2 | 2102 | 90 | 80 | 4 | 6 | 1.7834 | 23.0 |
| [Claude Opus 5.5 (Medium)](https://chessbench.org/models/claude-opus-5.5) | 6 | 1997 | 88 | 72 | 6 | 10 | 0.6702 | 14.0 |
| [Claude Sonnet 5.5 (xHigh)](https://chessbench.org/models/claude-sonnet-5.5) | 3 | 2089 | 74 | 65 | 0 | 9 | 2.2277 | 37.5 |
| [Claude Sonnet 5.5 (Medium)](https://chessbench.org/models/claude-sonnet-5.5) | 12 | 1641 | 74 | 36 | 8 | 30 | 0.4548 | 9.7 |
| [GPT-6 Astra (xHigh)](https://chessbench.org/models/gpt-6-astra) | 5 | 2022 | 122 | 104 | 1 | 17 | 1.4520 | 14.4 |
| [GPT-6 Astra (Medium)](https://chessbench.org/models/gpt-6-astra) | 10 | 1767 | 118 | 74 | 13 | 31 | 1.0280 | 6.5 |
| [Claude Fable 5.1 (xHigh)](https://chessbench.org/models/claude-fable-5.1) | 7 | 1891 | 120 | 85 | 7 | 28 | 8.3149 | 35.3 |
| [Claude Fable 5.1 (Medium)](https://chessbench.org/models/claude-fable-5.1) | 16 | 1602 | 116 | 51 | 21 | 44 | 2.2240 | 8.8 |
| [GPT-6.1 SOL (xHigh)](https://chessbench.org/models/gpt-6.1-sol) | 8 | 1840 | 70 | 48 | 3 | 19 | 0.3650 | 18.6 |
| [GPT-6.1 SOL (Medium)](https://chessbench.org/models/gpt-6.1-sol) | 13 | 1630 | 70 | 34 | 6 | 30 | 0.1665 | 7.1 |
| [Claude Haiku 5.5 (xHigh)](https://chessbench.org/models/claude-haiku-5.5) | 9 | 1796 | 70 | 44 | 6 | 20 | 0.2763 | 21.8 |
| [Claude Haiku 5.5 (Medium)](https://chessbench.org/models/claude-haiku-5.5) | 17 | 1584 | 70 | 27 | 10 | 33 | 0.0424 | 8.9 |
| [DeepSeek V4.1 Flash (1017) (Max)](https://chessbench.org/models/deepseek-flash-1017) | 11 | 1687 | 62 | 32 | 6 | 24 | 0.2006 | 36.9 |
| [DeepSeek V4.1 Flash (1017) (High)](https://chessbench.org/models/deepseek-flash-1017) | 15 | 1612 | 62 | 28 | 8 | 26 | 0.2105 | 39.0 |
| [GPT-6.1 Terra (xHigh)](https://chessbench.org/models/gpt-6.1-terra) | 14 | 1621 | 62 | 30 | 5 | 27 | 0.1090 | 5.7 |
| [GPT-6.1 Terra (Medium)](https://chessbench.org/models/gpt-6.1-terra) | 22 | 1451 | 62 | 19 | 11 | 32 | 0.0759 | 3.5 |
| [Grok 4.7 (xHigh)](https://chessbench.org/models/grok-4.7) | 18 | 1549 | 100 | 40 | 23 | 37 | 3.6145 | 48.5 |
| [Muse Spark 1.3 (Medium)](https://chessbench.org/models/muse-spark-1.3) | 19 | 1506 | 119 | 42 | 31 | 46 | 0.0291 | 19.1 |
| [Muse Spark 1.3 (xHigh)](https://chessbench.org/models/muse-spark-1.3) | 23 | 1426 | 118 | 42 | 21 | 55 | 0.0602 | 36.0 |
| [GLM 5.3 (High)](https://chessbench.org/models/glm-5.3) | 20 | 1501 | 120 | 45 | 17 | 58 | 2.1328 | 60.6 |
| [GLM 5.3 (Max)](https://chessbench.org/models/glm-5.3) | 21 | 1491 | 120 | 47 | 11 | 62 | 1.8645 | 60.2 |
| [GPT-6 Luna (xHigh)](https://chessbench.org/models/gpt-6-luna) | 24 | 1421 | 82 | 18 | 24 | 40 | 0.0536 | 29.1 |
| [GPT-6 Luna (Medium)](https://chessbench.org/models/gpt-6-luna) | 32 | 932 | 84 | 1 | 7 | 76 | 0.0066 | 3.9 |
| [MiniMax M3.1 Flash (xHigh)](https://chessbench.org/models/minimax-m3.1-flash) | 25 | 1386 | 62 | 15 | 6 | 41 | 0.2744 | 31.9 |
| [MiniMax M3.1 Flash (Medium)](https://chessbench.org/models/minimax-m3.1-flash) | 30 | 1189 | 62 | 7 | 5 | 50 | 0.1201 | 16.6 |
| [MiMo V2.6 Flash (Thinking)](https://chessbench.org/models/mimo-v2.6-flash) | 26 | 1345 | 96 | 20 | 20 | 56 | 0.0561 | 44.3 |
| [GLM 5.3 Flash (High)](https://chessbench.org/models/glm-5.3-flash) | 27 | 1327 | 121 | 24 | 29 | 68 | 0.1061 | 76.0 |
| [GLM 5.3 Flash (Max)](https://chessbench.org/models/glm-5.3-flash) | 28 | 1279 | 120 | 23 | 26 | 71 | 0.1032 | 102.7 |
| [MiniMax M3 (Thinking)](https://chessbench.org/models/minimax-m3) | 29 | 1261 | 119 | 15 | 33 | 71 | 0.5412 | 33.7 |
| [Qwen3.7 Plus (Thinking)](https://chessbench.org/models/qwen3.7-plus) | 31 | 1075 | 113 | 4 | 38 | 71 | 0.0724 | 8.3 |

Methodology: https://chessbench.org/methodology
Time includes model requests per accepted move. Cost uses fully priced completed games. Human games do not change these ratings. Terms: https://chessbench.org/terms