Benchmark Almanac

Game Arena Chess

Gemini 4 Carbon 1,646 leads since October 2026
Not in model cards

Language models play full chess games against each other with the board given as FEN and PGN text. Ratings come from many games between each pair.

The February 2nd, 2026 leaderboard is labeled v2 of the chess text task. Kaggle also runs Chess Openings (11 models) and a Chess Suite (19 models). 39 models on October 19th, 2026.

Made by
Kaggle and Google DeepMind
Released
August 2025
Kind
Puzzles and chess
Level
Expert
Scored
Ranked by matches
Answers
Agent task
Tasks
All public
Strongest model tested
Claude Fable 5.2 #1 on the capability index
Last model added
November 2026
Scores refreshed
November 12th, 2026
42
Models scored on its own leaderboard

Gemini 4 Carbon leads with 1,646

Best score per model from the benchmark's own leaderboard, read on November 12th, 2026. Where a model ran with several settings or harnesses, we keep its best.

42 models. Scroll sideways for the rest.

  1. 1,646 Gemini 4 Carbon
  2. 1,458 Gemini 3.8 Flash
  3. 1,435 GPT-6 Astra
  4. 1,397 Gemini 3.7 Flash
  5. 1,371 Gemini 4 Argon
  6. 1,342 Gemini 3.5 Flash
  7. 1,314 Gemini 3.1 Pro
  8. 1,313 Claude Fable 5.2
  9. 1,244 GPT-5.5
  10. 1,233 GPT-6.1 Sol
  11. 1,224 Gemini 3.6 Flash
  12. 1,213 Gemini 3 Flash
  13. 1,132 GPT-5.6 Sol
  14. 1,069 GPT-6.1 Terra
  15. 1,008 GPT-6 Sol
  16. 974 Claude Opus 5.5
  17. 925 o3
  18. 894 GPT-5.4
  19. 842 Grok 4.7
  20. 827 GPT-5.2
  21. 808 Grok 4.6
  22. 783 Claude Sonnet 5.5
  23. 761 Claude Fable 5.1
  24. 756 Grok 4
  25. 725 GPT-5.6 Terra
  26. 700 GPT-5.6 Luna
  27. 691 Claude Opus 5
  28. 666 Grok 4.20
  29. 645 Grok 4.5
  30. 621 Claude Opus 4.8
  31. 607 Grok 4.1 Fast
  32. 595 Grok 4.3
  33. 594 Claude Haiku 5.5
  34. 587 Claude Sonnet 5
  35. 539 GPT-5 mini
  36. 450 GPT-6 Luna
  37. 438 GPT-5.4 Mini
  38. 284 Claude Opus 4.7
  39. 98 Claude Sonnet 4.6
  40. 86 DeepSeek-V3.2
  41. 50 Claude Opus 4.6
  42. 0 Claude Sonnet 4.5
  1. 1 Gemini 4 Carbon 1,646
  2. 2 Gemini 3.8 Flash 1,458
  3. 3 GPT-6 Astra 1,435
  4. 4 Gemini 3.7 Flash 1,397
  5. 5 Gemini 4 Argon 1,371
  6. 6 Gemini 3.5 Flash 1,342
  7. 7 Gemini 3.1 Pro 1,314
  8. 8 Claude Fable 5.2 1,313
  9. 9 GPT-5.5 1,244
  10. 10 GPT-6.1 Sol 1,233
  11. 11 Gemini 3.6 Flash 1,224
  12. 12 Gemini 3 Flash 1,213
  13. 13 GPT-5.6 Sol 1,132
  14. 14 GPT-6.1 Terra 1,069
  15. 15 GPT-6 Sol 1,008
  16. 16 Claude Opus 5.5 974
  17. 17 o3 925
  18. 18 GPT-5.4 894
  19. 19 Grok 4.7 842
  20. 20 GPT-5.2 827
  21. 21 Grok 4.6 808
  22. 22 Claude Sonnet 5.5 783
  23. 23 Claude Fable 5.1 761
  24. 24 Grok 4 756
  25. 25 GPT-5.6 Terra 725
  26. 26 GPT-5.6 Luna 700
  27. 27 Claude Opus 5 691
  28. 28 Grok 4.20 reasoning 666
  29. 29 Grok 4.5 645
  30. 30 Claude Opus 4.8 621
  31. 31 Grok 4.1 Fast reasoning 607
  32. 32 Grok 4.3 595
  33. 33 Claude Haiku 5.5 594
  34. 34 Claude Sonnet 5 587
  35. 35 GPT-5 mini 539
  36. 36 GPT-6 Luna 450
  37. 37 GPT-5.4 Mini 438
  38. 38 Claude Opus 4.7 284
  39. 39 Claude Sonnet 4.6 98
  40. 40 DeepSeek-V3.2 86
  41. 41 Claude Opus 4.6 50
  42. 42 Claude Sonnet 4.5 0

Source: Official Game Arena Chess leaderboard

#ModelScoreSettingDateSource
1 Gemini 4 Carbon 1,646 Oct 2026 Official leaderboard
2 Gemini 3.8 Flash 1,458 Sep 2026 Official leaderboard
3 GPT-6 Astra 1,435 Sep 2026 Official leaderboard
4 Gemini 3.7 Flash 1,397 Aug 2026 Official leaderboard
5 Gemini 4 Argon 1,371 Sep 2026 Official leaderboard
6 Gemini 3.5 Flash 1,342 May 2026 Official leaderboard
7 Gemini 3.1 Pro 1,314 Feb 2026 Official leaderboard
8 Claude Fable 5.2 1,313 Nov 2026 Official leaderboard
9 GPT-5.5 1,244 Apr 2026 Official leaderboard
10 GPT-6.1 Sol 1,233 Sep 2026 Official leaderboard
11 Gemini 3.6 Flash 1,224 Jul 2026 Official leaderboard
12 Gemini 3 Flash 1,213 Dec 2025 Official leaderboard
13 GPT-5.6 Sol 1,132 Jul 2026 Official leaderboard
14 GPT-6.1 Terra 1,069 Oct 2026 Official leaderboard
15 GPT-6 Sol 1,008 Sep 2026 Official leaderboard
16 Claude Opus 5.5 974 Sep 2026 Official leaderboard
17 o3 925 Apr 2025 Official leaderboard
18 GPT-5.4 894 Mar 2026 Official leaderboard
19 Grok 4.7 842 Sep 2026 Official leaderboard
20 GPT-5.2 827 Dec 2025 Official leaderboard
21 Grok 4.6 808 Aug 2026 Official leaderboard
22 Claude Sonnet 5.5 783 Sep 2026 Official leaderboard
23 Claude Fable 5.1 761 Sep 2026 Official leaderboard
24 Grok 4 756 Jul 2025 Official leaderboard
25 GPT-5.6 Terra 725 Jul 2026 Official leaderboard
26 GPT-5.6 Luna 700 Jul 2026 Official leaderboard
27 Claude Opus 5 691 Jul 2026 Official leaderboard
28 Grok 4.20 666 reasoning Feb 2026 Official leaderboard
29 Grok 4.5 645 Jul 2026 Official leaderboard
30 Claude Opus 4.8 621 May 2026 Official leaderboard
31 Grok 4.1 Fast 607 reasoning Nov 2025 Official leaderboard
32 Grok 4.3 595 Apr 2026 Official leaderboard
33 Claude Haiku 5.5 594 Oct 2026 Official leaderboard
34 Claude Sonnet 5 587 Jun 2026 Official leaderboard
35 GPT-5 mini 539 Aug 2025 Official leaderboard
36 GPT-6 Luna 450 Sep 2026 Official leaderboard
37 GPT-5.4 Mini 438 Mar 2026 Official leaderboard
38 Claude Opus 4.7 284 Apr 2026 Official leaderboard
39 Claude Sonnet 4.6 98 Feb 2026 Official leaderboard
40 DeepSeek-V3.2 86 Dec 2025 Official leaderboard
41 Claude Opus 4.6 50 Feb 2026 Official leaderboard
42 Claude Sonnet 4.5 0 Sep 2025 Official leaderboard

7 models have held number one on the official leaderboard since its release in August 2025

Lab claims count from the day the card came out, independent runs from the model’s release.

OpenAI Google DeepMind
Own leaderboard 7 leaders
Gemini 4 Carbon 1,646

Source: Official Game Arena Chess leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

On the benchmark’s own leaderboard
0 500 1,000 1,500 2,000 2025 2026 2027 0 500 1,000 1,500 2,000 2025 2026 2027

Source: Official Game Arena Chess leaderboard

No lab model card reports it yet

We read the headline results of 522 launch posts from 52 labs, and none of them list this benchmark.

More in Games and long tasks

All 82 in Games and long tasks
  • ARC-AGI-3 In use Interactive game environments where agents must explore, infer goals and rules, and act efficiently without instructions. GPT-6 Astra 99.95%
  • Orak Fading LLM agents play 12 popular video games across most genres through an MCP interface. Each game scores in-game progress or outcomes, and the board ranks models by their average rank across the games. Gemini 2.5 Pro 3.90
  • MineDojo Fading Agents complete open-ended tasks in Minecraft, such as mining blocks and fighting mobs, from a large simulation suite. Programmatic checks score many tasks. Seed2.0 Pro 49%
  • TextQuests Only in system cards Agents play 25 classic interactive fiction games. The score is harmful actions per 100 steps, where lower is better. The board also reports game progress. Llama 4 Scout 7.00
  • GameWorld Only in system cards Multimodal game agents play 34 browser games across 170 tasks. Outcomes are checked from the game state. GPT-6.1 Terra 76.8%
  • Swarm Bench Only in system cards Models orchestrate agent swarms on complex tasks that benefit from coordinated decomposition and parallel execution. Kimi K3 76.3%