Benchmark Almanac

Game Arena Clobber

GPT-6 Astra 847 leads since Sep 2026 Not in model cards

Language models play Clobber, where every move captures an adjacent enemy stone. The player who makes the last legal move wins.

Kaggle published the leaderboard on 2026-06-25 with 36 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
36
Models scored on its own leaderboard

GPT-6 Astra leads with 847

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.

36 models. Scroll sideways for the rest.

  1. 847 GPT-6 Astra
  2. 720 Gemini 4 Argon
  3. 655 GPT-6.1 Sol
  4. 566 Claude Opus 5
  5. 517 Claude Fable 5.1
  6. 486 Claude Opus 5.5
  7. 478 Gemini 3.8 Flash
  8. 475 Gemini 3.1 Pro
  9. 452 GPT-5.4
  10. 410 GPT-5.5
  11. 390 Gemini 3.7 Flash
  12. 348 GPT-5.2
  13. 340 GPT-5.6 Sol
  14. 334 Grok 4.7
  15. 326 Claude Opus 4.6
  16. 314 GPT-6 Sol
  17. 303 Claude Opus 4.8
  18. 295 Gemini 3.6 Flash
  19. 265 Claude Sonnet 4.6
  20. 248 Claude Sonnet 5
  21. 225 Claude Sonnet 4.5
  22. 209 Gemini 3.5 Flash
  23. 168 Claude Sonnet 5.5
  24. 156 Grok 4.6
  25. 154 GPT-5.6 Terra
  26. 152 Grok 4.5
  27. 145 GPT-5.4 Mini
  28. 141 o3
  29. 128 Claude Haiku 5.5
  30. 107 Grok 4.20
  31. 60 GPT-6 Luna
  32. 41 GPT-5 mini
  33. 35 GPT-5.6 Luna
  34. 25 Claude Opus 4.7
  35. 21 Gemini 3 Flash
  36. 0 Grok 4.3
  1. 1 GPT-6 Astra 847
  2. 2 Gemini 4 Argon 720
  3. 3 GPT-6.1 Sol 655
  4. 4 Claude Opus 5 566
  5. 5 Claude Fable 5.1 517
  6. 6 Claude Opus 5.5 486
  7. 7 Gemini 3.8 Flash 478
  8. 8 Gemini 3.1 Pro 475
  9. 9 GPT-5.4 452
  10. 10 GPT-5.5 410
  11. 11 Gemini 3.7 Flash 390
  12. 12 GPT-5.2 348
  13. 13 GPT-5.6 Sol 340
  14. 14 Grok 4.7 334
  15. 15 Claude Opus 4.6 326
  16. 16 GPT-6 Sol 314
  17. 17 Claude Opus 4.8 303
  18. 18 Gemini 3.6 Flash 295
  19. 19 Claude Sonnet 4.6 265
  20. 20 Claude Sonnet 5 248
  21. 21 Claude Sonnet 4.5 225
  22. 22 Gemini 3.5 Flash 209
  23. 23 Claude Sonnet 5.5 168
  24. 24 Grok 4.6 156
  25. 25 GPT-5.6 Terra 154
  26. 26 Grok 4.5 152
  27. 27 GPT-5.4 Mini 145
  28. 28 o3 141
  29. 29 Claude Haiku 5.5 128
  30. 30 Grok 4.20 reasoning 107
  31. 31 GPT-6 Luna 60
  32. 32 GPT-5 mini 41
  33. 33 GPT-5.6 Luna 35
  34. 34 Claude Opus 4.7 25
  35. 35 Gemini 3 Flash 21
  36. 36 Grok 4.3 0
#ModelScoreSettingDateSource
1 GPT-6 Astra 847 Sep 2026 Official leaderboard
2 Gemini 4 Argon 720 Sep 2026 Official leaderboard
3 GPT-6.1 Sol 655 Sep 2026 Official leaderboard
4 Claude Opus 5 566 Jul 2026 Official leaderboard
5 Claude Fable 5.1 517 Sep 2026 Official leaderboard
6 Claude Opus 5.5 486 Sep 2026 Official leaderboard
7 Gemini 3.8 Flash 478 Sep 2026 Official leaderboard
8 Gemini 3.1 Pro 475 Feb 2026 Official leaderboard
9 GPT-5.4 452 Mar 2026 Official leaderboard
10 GPT-5.5 410 Apr 2026 Official leaderboard
11 Gemini 3.7 Flash 390 Aug 2026 Official leaderboard
12 GPT-5.2 348 Dec 2025 Official leaderboard
13 GPT-5.6 Sol 340 Jul 2026 Official leaderboard
14 Grok 4.7 334 Sep 2026 Official leaderboard
15 Claude Opus 4.6 326 Feb 2026 Official leaderboard
16 GPT-6 Sol 314 Sep 2026 Official leaderboard
17 Claude Opus 4.8 303 May 2026 Official leaderboard
18 Gemini 3.6 Flash 295 Jul 2026 Official leaderboard
19 Claude Sonnet 4.6 265 Feb 2026 Official leaderboard
20 Claude Sonnet 5 248 Jun 2026 Official leaderboard
21 Claude Sonnet 4.5 225 Sep 2025 Official leaderboard
22 Gemini 3.5 Flash 209 May 2026 Official leaderboard
23 Claude Sonnet 5.5 168 Sep 2026 Official leaderboard
24 Grok 4.6 156 Aug 2026 Official leaderboard
25 GPT-5.6 Terra 154 Jul 2026 Official leaderboard
26 Grok 4.5 152 Jul 2026 Official leaderboard
27 GPT-5.4 Mini 145 Mar 2026 Official leaderboard
28 o3 141 Apr 2025 Official leaderboard
29 Claude Haiku 5.5 128 Oct 2026 Official leaderboard
30 Grok 4.20 107 reasoning Feb 2026 Official leaderboard
31 GPT-6 Luna 60 Sep 2026 Official leaderboard
32 GPT-5 mini 41 Aug 2025 Official leaderboard
33 GPT-5.6 Luna 35 Jul 2026 Official leaderboard
34 Claude Opus 4.7 25 Apr 2026 Official leaderboard
35 Gemini 3 Flash 21 Dec 2025 Official leaderboard
36 Grok 4.3 0 Apr 2026 Official leaderboard

3 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.

Google DeepMind Anthropic OpenAI
Own leaderboard 3 leaders
GPT-6 Astra 847

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
0 200 400 600 800 1,000 2025 2026 0 200 400 600 800 1,000 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