Benchmark Almanac

Game Arena Dots and Boxes

GPT-6 Astra 1,219 leads since Sep 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 2026-06-17 with 37 models on 2026-11-01. Gameplay logs are a public Kaggle dataset. Part of the Kaggle Game Arena suite.

Made by Kaggle and Google DeepMind Released Jun 2026 Puzzles and chess All tasks public Last model added Oct 2026 Scores refreshed 2 Nov 2026
Level
Everyday Most adults, without training
Scored
Ranked by matches Models play against each other, and the results become ratings.
Answers
Agent task Many steps in a terminal, browser, desktop, app, game or robot.
Kaggle/kaggle-environments 459 stars Dataset Leaderboard kaggle.com
37
Models scored on its own leaderboard

GPT-6 Astra leads with 1,219

Best score per model from the benchmark's own leaderboard, read on 2 Nov 2026. Where a model ran with several settings or harnesses, we keep its best.

37 models. Scroll sideways for the rest.

  1. 1,219 GPT-6 Astra
  2. 1,172 Claude Opus 5.5
  3. 1,134 GPT-6.1 Sol
  4. 1,117 Gemini 4 Argon
  5. 1,116 Claude Fable 5.1
  6. 1,077 Claude Opus 5
  7. 1,069 Claude Sonnet 5.5
  8. 1,028 Gemini 3.8 Flash
  9. 1,018 Claude Opus 4.8
  10. 990 GPT-5.5
  11. 976 Gemini 3.1 Pro
  12. 966 Claude Haiku 5.5
  13. 950 GPT-5.4
  14. 948 Grok 4.6
  15. 945 Claude Sonnet 5
  16. 933 GPT-5.6 Sol
  17. 927 Grok 4.7
  18. 908 Gemini 3.7 Flash
  19. 904 Gemini 3.5 Flash
  20. 879 GPT-5.2
  21. 875 Grok 4.5
  22. 872 Claude Opus 4.6
  23. 817 GPT-6 Sol
  24. 805 Claude Sonnet 4.6
  25. 787 Gemini 3.6 Flash
  26. 757 Gemini 3 Flash
  27. 708 GPT-5 mini
  28. 625 Grok 4.20
  29. 621 GPT-5.6 Terra
  30. 612 GPT-5.4 Mini
  31. 611 GPT-6 Luna
  32. 594 o3
  33. 529 Claude Sonnet 4.5
  34. 442 Grok 4.3
  35. 438 GPT-5.6 Luna
  36. 387 Claude Opus 4.7
  37. 0 DeepSeek-V3.2
  1. 1 GPT-6 Astra 1,219
  2. 2 Claude Opus 5.5 1,172
  3. 3 GPT-6.1 Sol 1,134
  4. 4 Gemini 4 Argon 1,117
  5. 5 Claude Fable 5.1 1,116
  6. 6 Claude Opus 5 1,077
  7. 7 Claude Sonnet 5.5 1,069
  8. 8 Gemini 3.8 Flash 1,028
  9. 9 Claude Opus 4.8 1,018
  10. 10 GPT-5.5 990
  11. 11 Gemini 3.1 Pro 976
  12. 12 Claude Haiku 5.5 966
  13. 13 GPT-5.4 950
  14. 14 Grok 4.6 948
  15. 15 Claude Sonnet 5 945
  16. 16 GPT-5.6 Sol 933
  17. 17 Grok 4.7 927
  18. 18 Gemini 3.7 Flash 908
  19. 19 Gemini 3.5 Flash 904
  20. 20 GPT-5.2 879
  21. 21 Grok 4.5 875
  22. 22 Claude Opus 4.6 872
  23. 23 GPT-6 Sol 817
  24. 24 Claude Sonnet 4.6 805
  25. 25 Gemini 3.6 Flash 787
  26. 26 Gemini 3 Flash 757
  27. 27 GPT-5 mini 708
  28. 28 Grok 4.20 reasoning 625
  29. 29 GPT-5.6 Terra 621
  30. 30 GPT-5.4 Mini 612
  31. 31 GPT-6 Luna 611
  32. 32 o3 594
  33. 33 Claude Sonnet 4.5 529
  34. 34 Grok 4.3 442
  35. 35 GPT-5.6 Luna 438
  36. 36 Claude Opus 4.7 387
  37. 37 DeepSeek-V3.2 0
#ModelScoreSettingDateSource
1 GPT-6 Astra 1,219 Sep 2026 Official leaderboard
2 Claude Opus 5.5 1,172 Sep 2026 Official leaderboard
3 GPT-6.1 Sol 1,134 Sep 2026 Official leaderboard
4 Gemini 4 Argon 1,117 Sep 2026 Official leaderboard
5 Claude Fable 5.1 1,116 Sep 2026 Official leaderboard
6 Claude Opus 5 1,077 Jul 2026 Official leaderboard
7 Claude Sonnet 5.5 1,069 Sep 2026 Official leaderboard
8 Gemini 3.8 Flash 1,028 Sep 2026 Official leaderboard
9 Claude Opus 4.8 1,018 May 2026 Official leaderboard
10 GPT-5.5 990 Apr 2026 Official leaderboard
11 Gemini 3.1 Pro 976 Feb 2026 Official leaderboard
12 Claude Haiku 5.5 966 Oct 2026 Official leaderboard
13 GPT-5.4 950 Mar 2026 Official leaderboard
14 Grok 4.6 948 Aug 2026 Official leaderboard
15 Claude Sonnet 5 945 Jun 2026 Official leaderboard
16 GPT-5.6 Sol 933 Jul 2026 Official leaderboard
17 Grok 4.7 927 Sep 2026 Official leaderboard
18 Gemini 3.7 Flash 908 Aug 2026 Official leaderboard
19 Gemini 3.5 Flash 904 May 2026 Official leaderboard
20 GPT-5.2 879 Dec 2025 Official leaderboard
21 Grok 4.5 875 Jul 2026 Official leaderboard
22 Claude Opus 4.6 872 Feb 2026 Official leaderboard
23 GPT-6 Sol 817 Sep 2026 Official leaderboard
24 Claude Sonnet 4.6 805 Feb 2026 Official leaderboard
25 Gemini 3.6 Flash 787 Jul 2026 Official leaderboard
26 Gemini 3 Flash 757 Dec 2025 Official leaderboard
27 GPT-5 mini 708 Aug 2025 Official leaderboard
28 Grok 4.20 625 reasoning Feb 2026 Official leaderboard
29 GPT-5.6 Terra 621 Jul 2026 Official leaderboard
30 GPT-5.4 Mini 612 Mar 2026 Official leaderboard
31 GPT-6 Luna 611 Sep 2026 Official leaderboard
32 o3 594 Apr 2025 Official leaderboard
33 Claude Sonnet 4.5 529 Sep 2025 Official leaderboard
34 Grok 4.3 442 Apr 2026 Official leaderboard
35 GPT-5.6 Luna 438 Jul 2026 Official leaderboard
36 Claude Opus 4.7 387 Apr 2026 Official leaderboard
37 DeepSeek-V3.2 0 Dec 2025 Official leaderboard

4 models have held number one since its release in Jun 2026

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

Anthropic OpenAI
Own leaderboard 4 leaders
GPT-6 Astra 1,219

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 0 250 500 750 1,000 1,250 2025 2026

No lab model card reports it yet

We read the headline results of 479 launch posts from 51 labs, and none of them list this benchmark.

More in Games and long tasks

All 68 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 100%
  • 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%
  • Orak Fading LLM agents play 12 popular video games across most genres through an MCP interface. Scores come from in-game progress or outcomes per title. Nemotron 3 Nano 4B 22.9
  • GameWorld Only in system cards Multimodal game agents play 34 browser games across 170 tasks. Outcomes are checked from the game state. Gemini 3 Flash 41.9%
  • 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%
  • TextQuests Only in system cards Agents play 25 classic interactive fiction games. Measures game progress and harmful actions per 100 steps. Muse Spark 2.40