Benchmark Almanac

Game Arena Nine Men's Morris

Gemini 4 Carbon 1,190 leads since October 2026
Not in model cards

Language models play Nine Men's Morris. Forming a row of three, called a mill, lets a player remove one of the opponent's pieces.

Kaggle published the leaderboard on August 4th, 2026 with 36 models on November 1st, 2026. Gameplay logs are a public Kaggle dataset. Part of the Kaggle Game Arena suite.

Made by
Kaggle and Google DeepMind
Released
August 2026
Kind
Puzzles and chess
Level
Everyday
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
39
Models scored on its own leaderboard

Gemini 4 Carbon leads with 1,190

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.

39 models. Scroll sideways for the rest.

  1. 1,190 Gemini 4 Carbon
  2. 1,177 Claude Fable 5.2
  3. 1,010 Claude Opus 5
  4. 1,001 Claude Fable 5.1
  5. 981 Gemini 4 Argon
  6. 960 Claude Opus 5.5
  7. 912 Claude Sonnet 5.5
  8. 896 Grok 4.7
  9. 819 Claude Opus 4.8
  10. 794 Claude Haiku 5.5
  11. 788 Gemini 3.8 Flash
  12. 787 Gemini 3.1 Pro
  13. 780 GPT-6.1 Sol
  14. 772 GPT-6 Astra
  15. 750 Claude Sonnet 5
  16. 711 Gemini 3.7 Flash
  17. 686 Grok 4.6
  18. 665 GPT-6.1 Terra
  19. 637 GPT-5.6 Sol
  20. 621 GPT-5.5
  21. 610 GPT-5.4
  22. 595 Gemini 3.6 Flash
  23. 585 Gemini 3 Flash
  24. 578 GPT-6 Sol
  25. 565 o3
  26. 560 Claude Opus 4.7
  27. 530 Grok 4.5
  28. 527 Claude Opus 4.6
  29. 525 Gemini 3.5 Flash
  30. 512 GPT-5.2
  31. 428 Claude Sonnet 4.6
  32. 378 Grok 4.20
  33. 359 Claude Sonnet 4.5
  34. 292 GPT-5 mini
  35. 262 GPT-5.6 Terra
  36. 213 GPT-5.4 Mini
  37. 33 GPT-5.6 Luna
  38. 11 Grok 4.3
  39. 0 GPT-6 Luna
  1. 1 Gemini 4 Carbon 1,190
  2. 2 Claude Fable 5.2 1,177
  3. 3 Claude Opus 5 1,010
  4. 4 Claude Fable 5.1 1,001
  5. 5 Gemini 4 Argon 981
  6. 6 Claude Opus 5.5 960
  7. 7 Claude Sonnet 5.5 912
  8. 8 Grok 4.7 896
  9. 9 Claude Opus 4.8 819
  10. 10 Claude Haiku 5.5 794
  11. 11 Gemini 3.8 Flash 788
  12. 12 Gemini 3.1 Pro 787
  13. 13 GPT-6.1 Sol 780
  14. 14 GPT-6 Astra 772
  15. 15 Claude Sonnet 5 750
  16. 16 Gemini 3.7 Flash 711
  17. 17 Grok 4.6 686
  18. 18 GPT-6.1 Terra 665
  19. 19 GPT-5.6 Sol 637
  20. 20 GPT-5.5 621
  21. 21 GPT-5.4 610
  22. 22 Gemini 3.6 Flash 595
  23. 23 Gemini 3 Flash 585
  24. 24 GPT-6 Sol 578
  25. 25 o3 565
  26. 26 Claude Opus 4.7 560
  27. 27 Grok 4.5 530
  28. 28 Claude Opus 4.6 527
  29. 29 Gemini 3.5 Flash 525
  30. 30 GPT-5.2 512
  31. 31 Claude Sonnet 4.6 428
  32. 32 Grok 4.20 reasoning 378
  33. 33 Claude Sonnet 4.5 359
  34. 34 GPT-5 mini 292
  35. 35 GPT-5.6 Terra 262
  36. 36 GPT-5.4 Mini 213
  37. 37 GPT-5.6 Luna 33
  38. 38 Grok 4.3 11
  39. 39 GPT-6 Luna 0

Source: Official Game Arena Nine Men's Morris leaderboard

#ModelScoreSettingDateSource
1 Gemini 4 Carbon 1,190 Oct 2026 Official leaderboard
2 Claude Fable 5.2 1,177 Nov 2026 Official leaderboard
3 Claude Opus 5 1,010 Jul 2026 Official leaderboard
4 Claude Fable 5.1 1,001 Sep 2026 Official leaderboard
5 Gemini 4 Argon 981 Sep 2026 Official leaderboard
6 Claude Opus 5.5 960 Sep 2026 Official leaderboard
7 Claude Sonnet 5.5 912 Sep 2026 Official leaderboard
8 Grok 4.7 896 Sep 2026 Official leaderboard
9 Claude Opus 4.8 819 May 2026 Official leaderboard
10 Claude Haiku 5.5 794 Oct 2026 Official leaderboard
11 Gemini 3.8 Flash 788 Sep 2026 Official leaderboard
12 Gemini 3.1 Pro 787 Feb 2026 Official leaderboard
13 GPT-6.1 Sol 780 Sep 2026 Official leaderboard
14 GPT-6 Astra 772 Sep 2026 Official leaderboard
15 Claude Sonnet 5 750 Jun 2026 Official leaderboard
16 Gemini 3.7 Flash 711 Aug 2026 Official leaderboard
17 Grok 4.6 686 Aug 2026 Official leaderboard
18 GPT-6.1 Terra 665 Oct 2026 Official leaderboard
19 GPT-5.6 Sol 637 Jul 2026 Official leaderboard
20 GPT-5.5 621 Apr 2026 Official leaderboard
21 GPT-5.4 610 Mar 2026 Official leaderboard
22 Gemini 3.6 Flash 595 Jul 2026 Official leaderboard
23 Gemini 3 Flash 585 Dec 2025 Official leaderboard
24 GPT-6 Sol 578 Sep 2026 Official leaderboard
25 o3 565 Apr 2025 Official leaderboard
26 Claude Opus 4.7 560 Apr 2026 Official leaderboard
27 Grok 4.5 530 Jul 2026 Official leaderboard
28 Claude Opus 4.6 527 Feb 2026 Official leaderboard
29 Gemini 3.5 Flash 525 May 2026 Official leaderboard
30 GPT-5.2 512 Dec 2025 Official leaderboard
31 Claude Sonnet 4.6 428 Feb 2026 Official leaderboard
32 Grok 4.20 378 reasoning Feb 2026 Official leaderboard
33 Claude Sonnet 4.5 359 Sep 2025 Official leaderboard
34 GPT-5 mini 292 Aug 2025 Official leaderboard
35 GPT-5.6 Terra 262 Jul 2026 Official leaderboard
36 GPT-5.4 Mini 213 Mar 2026 Official leaderboard
37 GPT-5.6 Luna 33 Jul 2026 Official leaderboard
38 Grok 4.3 11 Apr 2026 Official leaderboard
39 GPT-6 Luna 0 Sep 2026 Official leaderboard

2 models have held number one on the official leaderboard since its release in August 2026

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

Anthropic Google DeepMind
Own leaderboard 2 leaders
Gemini 4 Carbon 1,190

Source: Official Game Arena Nine Men's Morris 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
0 250 500 750 1,000 1,250 2025 2026 2027 0 250 500 750 1,000 1,250 2025 2026 2027

Source: Official Game Arena Nine Men's Morris 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%