Benchmark Almanac

Game Arena Dots and Boxes

Claude Fable 5.2 1,393 leads since November 2026
Not in model cards

Language models take turns drawing lines between dots on a grid. Closing the fourth side of a box claims it.

Kaggle published the leaderboard on June 17th, 2026 with 37 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
June 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
40
Models scored on its own leaderboard

Claude Fable 5.2 leads with 1,393

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.

40 models. Scroll sideways for the rest.

  1. 1,393 Claude Fable 5.2
  2. 1,202 GPT-6 Astra
  3. 1,154 Claude Opus 5.5
  4. 1,120 GPT-6.1 Sol
  5. 1,099 Gemini 4 Argon
  6. 1,096 Claude Fable 5.1
  7. 1,066 Claude Opus 5
  8. 1,055 Claude Sonnet 5.5
  9. 1,018 Gemini 3.8 Flash
  10. 1,005 Claude Opus 4.8
  11. 983 GPT-6.1 Terra
  12. 974 GPT-5.5
  13. 966 Gemini 3.1 Pro
  14. 958 Claude Haiku 5.5
  15. 941 GPT-5.4
  16. 935 Grok 4.6
  17. 931 Claude Sonnet 5
  18. 923 GPT-5.6 Sol
  19. 917 Grok 4.7
  20. 901 Gemini 3.7 Flash
  21. 897 Gemini 3.5 Flash
  22. 878 GPT-5.2
  23. 865 Claude Opus 4.6
  24. 857 Grok 4.5
  25. 824 GPT-6 Sol
  26. 801 Claude Sonnet 4.6
  27. 789 Gemini 3.6 Flash
  28. 755 Gemini 3 Flash
  29. 708 GPT-5 mini
  30. 629 Grok 4.20
  31. 607 GPT-5.4 Mini
  32. 603 GPT-5.6 Terra
  33. 599 o3
  34. 583 GPT-6 Luna
  35. 570 Claude Opus 4.7
  36. 559 Grok 4.3
  37. 544 Claude Sonnet 4.5
  38. 502 GPT-5.6 Luna
  39. 487 Gemini 4 Carbon
  40. 0 DeepSeek-V3.2
  1. 1 Claude Fable 5.2 1,393
  2. 2 GPT-6 Astra 1,202
  3. 3 Claude Opus 5.5 1,154
  4. 4 GPT-6.1 Sol 1,120
  5. 5 Gemini 4 Argon 1,099
  6. 6 Claude Fable 5.1 1,096
  7. 7 Claude Opus 5 1,066
  8. 8 Claude Sonnet 5.5 1,055
  9. 9 Gemini 3.8 Flash 1,018
  10. 10 Claude Opus 4.8 1,005
  11. 11 GPT-6.1 Terra 983
  12. 12 GPT-5.5 974
  13. 13 Gemini 3.1 Pro 966
  14. 14 Claude Haiku 5.5 958
  15. 15 GPT-5.4 941
  16. 16 Grok 4.6 935
  17. 17 Claude Sonnet 5 931
  18. 18 GPT-5.6 Sol 923
  19. 19 Grok 4.7 917
  20. 20 Gemini 3.7 Flash 901
  21. 21 Gemini 3.5 Flash 897
  22. 22 GPT-5.2 878
  23. 23 Claude Opus 4.6 865
  24. 24 Grok 4.5 857
  25. 25 GPT-6 Sol 824
  26. 26 Claude Sonnet 4.6 801
  27. 27 Gemini 3.6 Flash 789
  28. 28 Gemini 3 Flash 755
  29. 29 GPT-5 mini 708
  30. 30 Grok 4.20 reasoning 629
  31. 31 GPT-5.4 Mini 607
  32. 32 GPT-5.6 Terra 603
  33. 33 o3 599
  34. 34 GPT-6 Luna 583
  35. 35 Claude Opus 4.7 570
  36. 36 Grok 4.3 559
  37. 37 Claude Sonnet 4.5 544
  38. 38 GPT-5.6 Luna 502
  39. 39 Gemini 4 Carbon 487
  40. 40 DeepSeek-V3.2 0

Source: Official Game Arena Dots and Boxes leaderboard

#ModelScoreSettingDateSource
1 Claude Fable 5.2 1,393 Nov 2026 Official leaderboard
2 GPT-6 Astra 1,202 Sep 2026 Official leaderboard
3 Claude Opus 5.5 1,154 Sep 2026 Official leaderboard
4 GPT-6.1 Sol 1,120 Sep 2026 Official leaderboard
5 Gemini 4 Argon 1,099 Sep 2026 Official leaderboard
6 Claude Fable 5.1 1,096 Sep 2026 Official leaderboard
7 Claude Opus 5 1,066 Jul 2026 Official leaderboard
8 Claude Sonnet 5.5 1,055 Sep 2026 Official leaderboard
9 Gemini 3.8 Flash 1,018 Sep 2026 Official leaderboard
10 Claude Opus 4.8 1,005 May 2026 Official leaderboard
11 GPT-6.1 Terra 983 Oct 2026 Official leaderboard
12 GPT-5.5 974 Apr 2026 Official leaderboard
13 Gemini 3.1 Pro 966 Feb 2026 Official leaderboard
14 Claude Haiku 5.5 958 Oct 2026 Official leaderboard
15 GPT-5.4 941 Mar 2026 Official leaderboard
16 Grok 4.6 935 Aug 2026 Official leaderboard
17 Claude Sonnet 5 931 Jun 2026 Official leaderboard
18 GPT-5.6 Sol 923 Jul 2026 Official leaderboard
19 Grok 4.7 917 Sep 2026 Official leaderboard
20 Gemini 3.7 Flash 901 Aug 2026 Official leaderboard
21 Gemini 3.5 Flash 897 May 2026 Official leaderboard
22 GPT-5.2 878 Dec 2025 Official leaderboard
23 Claude Opus 4.6 865 Feb 2026 Official leaderboard
24 Grok 4.5 857 Jul 2026 Official leaderboard
25 GPT-6 Sol 824 Sep 2026 Official leaderboard
26 Claude Sonnet 4.6 801 Feb 2026 Official leaderboard
27 Gemini 3.6 Flash 789 Jul 2026 Official leaderboard
28 Gemini 3 Flash 755 Dec 2025 Official leaderboard
29 GPT-5 mini 708 Aug 2025 Official leaderboard
30 Grok 4.20 629 reasoning Feb 2026 Official leaderboard
31 GPT-5.4 Mini 607 Mar 2026 Official leaderboard
32 GPT-5.6 Terra 603 Jul 2026 Official leaderboard
33 o3 599 Apr 2025 Official leaderboard
34 GPT-6 Luna 583 Sep 2026 Official leaderboard
35 Claude Opus 4.7 570 Apr 2026 Official leaderboard
36 Grok 4.3 559 Apr 2026 Official leaderboard
37 Claude Sonnet 4.5 544 Sep 2025 Official leaderboard
38 GPT-5.6 Luna 502 Jul 2026 Official leaderboard
39 Gemini 4 Carbon 487 Oct 2026 Official leaderboard
40 DeepSeek-V3.2 0 Dec 2025 Official leaderboard

5 models have held number one on the official leaderboard since its release in June 2026

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

Anthropic OpenAI
Own leaderboard 5 leaders
Claude Fable 5.2 1,393

Source: Official Game Arena Dots and Boxes 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 2025 2026 2027 0 500 1,000 1,500 2025 2026 2027

Source: Official Game Arena Dots and Boxes 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%