Benchmark Almanac

Game Arena Go

GPT-6 Astra 1,762 leads since Sep 2026 Not in model cards

Language models play Go against each other through the Game Arena text harness. Ratings come from head-to-head games.

Google promised Go at the 2025-08 launch. The leaderboard was published on 2026-09-22 with 36 models.

Made by Kaggle and Google DeepMind Released Sep 2026 Puzzles and chess All tasks public Last model added Oct 2026 Scores refreshed 2 Nov 2026
Level
Expert Specialists with years of training, often over hours of work
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.
google-deepmind/game_arena 117 stars Leaderboard kaggle.com
36
Models scored on its own leaderboard

GPT-6 Astra leads with 1,762

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. 1,762 GPT-6 Astra
  2. 1,720 Gemini 4 Argon
  3. 1,615 Claude Fable 5.1
  4. 1,560 Claude Opus 5
  5. 1,554 Claude Opus 5.5
  6. 1,450 Claude Sonnet 5.5
  7. 1,441 GPT-6.1 Sol
  8. 1,217 Gemini 3.8 Flash
  9. 1,149 Gemini 3.1 Pro
  10. 1,126 Claude Sonnet 5
  11. 1,097 Claude Opus 4.8
  12. 1,082 Claude Haiku 5.5
  13. 1,016 Grok 4.6
  14. 1,011 GPT-5.6 Sol
  15. 1,007 Grok 4.7
  16. 1,006 Gemini 3.7 Flash
  17. 998 GPT-5.5
  18. 910 Grok 4.5
  19. 905 GPT-6 Sol
  20. 854 GPT-5.4
  21. 758 Gemini 3.5 Flash
  22. 719 Gemini 3.6 Flash
  23. 568 Claude Opus 4.6
  24. 557 GPT-5.2
  25. 554 o3
  26. 512 GPT-5.6 Terra
  27. 508 Claude Opus 4.7
  28. 497 Gemini 3 Flash
  29. 476 GPT-6 Luna
  30. 470 Grok 4.20
  31. 446 GPT-5.6 Luna
  32. 325 Claude Sonnet 4.5
  33. 296 Claude Sonnet 4.6
  34. 130 Grok 4.3
  35. 121 GPT-5.4 Mini
  36. 0 GPT-5 mini
  1. 1 GPT-6 Astra 1,762
  2. 2 Gemini 4 Argon 1,720
  3. 3 Claude Fable 5.1 1,615
  4. 4 Claude Opus 5 1,560
  5. 5 Claude Opus 5.5 1,554
  6. 6 Claude Sonnet 5.5 1,450
  7. 7 GPT-6.1 Sol 1,441
  8. 8 Gemini 3.8 Flash 1,217
  9. 9 Gemini 3.1 Pro 1,149
  10. 10 Claude Sonnet 5 1,126
  11. 11 Claude Opus 4.8 1,097
  12. 12 Claude Haiku 5.5 1,082
  13. 13 Grok 4.6 1,016
  14. 14 GPT-5.6 Sol 1,011
  15. 15 Grok 4.7 1,007
  16. 16 Gemini 3.7 Flash 1,006
  17. 17 GPT-5.5 998
  18. 18 Grok 4.5 910
  19. 19 GPT-6 Sol 905
  20. 20 GPT-5.4 854
  21. 21 Gemini 3.5 Flash 758
  22. 22 Gemini 3.6 Flash 719
  23. 23 Claude Opus 4.6 568
  24. 24 GPT-5.2 557
  25. 25 o3 554
  26. 26 GPT-5.6 Terra 512
  27. 27 Claude Opus 4.7 508
  28. 28 Gemini 3 Flash 497
  29. 29 GPT-6 Luna 476
  30. 30 Grok 4.20 reasoning 470
  31. 31 GPT-5.6 Luna 446
  32. 32 Claude Sonnet 4.5 325
  33. 33 Claude Sonnet 4.6 296
  34. 34 Grok 4.3 130
  35. 35 GPT-5.4 Mini 121
  36. 36 GPT-5 mini 0
#ModelScoreSettingDateSource
1 GPT-6 Astra 1,762 Sep 2026 Official leaderboard
2 Gemini 4 Argon 1,720 Sep 2026 Official leaderboard
3 Claude Fable 5.1 1,615 Sep 2026 Official leaderboard
4 Claude Opus 5 1,560 Jul 2026 Official leaderboard
5 Claude Opus 5.5 1,554 Sep 2026 Official leaderboard
6 Claude Sonnet 5.5 1,450 Sep 2026 Official leaderboard
7 GPT-6.1 Sol 1,441 Sep 2026 Official leaderboard
8 Gemini 3.8 Flash 1,217 Sep 2026 Official leaderboard
9 Gemini 3.1 Pro 1,149 Feb 2026 Official leaderboard
10 Claude Sonnet 5 1,126 Jun 2026 Official leaderboard
11 Claude Opus 4.8 1,097 May 2026 Official leaderboard
12 Claude Haiku 5.5 1,082 Oct 2026 Official leaderboard
13 Grok 4.6 1,016 Aug 2026 Official leaderboard
14 GPT-5.6 Sol 1,011 Jul 2026 Official leaderboard
15 Grok 4.7 1,007 Sep 2026 Official leaderboard
16 Gemini 3.7 Flash 1,006 Aug 2026 Official leaderboard
17 GPT-5.5 998 Apr 2026 Official leaderboard
18 Grok 4.5 910 Jul 2026 Official leaderboard
19 GPT-6 Sol 905 Sep 2026 Official leaderboard
20 GPT-5.4 854 Mar 2026 Official leaderboard
21 Gemini 3.5 Flash 758 May 2026 Official leaderboard
22 Gemini 3.6 Flash 719 Jul 2026 Official leaderboard
23 Claude Opus 4.6 568 Feb 2026 Official leaderboard
24 GPT-5.2 557 Dec 2025 Official leaderboard
25 o3 554 Apr 2025 Official leaderboard
26 GPT-5.6 Terra 512 Jul 2026 Official leaderboard
27 Claude Opus 4.7 508 Apr 2026 Official leaderboard
28 Gemini 3 Flash 497 Dec 2025 Official leaderboard
29 GPT-6 Luna 476 Sep 2026 Official leaderboard
30 Grok 4.20 470 reasoning Feb 2026 Official leaderboard
31 GPT-5.6 Luna 446 Jul 2026 Official leaderboard
32 Claude Sonnet 4.5 325 Sep 2025 Official leaderboard
33 Claude Sonnet 4.6 296 Feb 2026 Official leaderboard
34 Grok 4.3 130 Apr 2026 Official leaderboard
35 GPT-5.4 Mini 121 Mar 2026 Official leaderboard
36 GPT-5 mini 0 Aug 2025 Official leaderboard

2 models have held number one since its release in Sep 2026

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

Anthropic OpenAI
Own leaderboard 2 leaders
GPT-6 Astra 1,762

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 0 500 1,000 1,500 2,000 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