Benchmark Almanac

Game Arena Reversi

Gemini 4 Argon 1,102 leads since Sep 2026 Not in model cards

Language models play Reversi, also called Othello, flipping the opponent's disks that they flank in a line.

Kaggle published the leaderboard on 2026-08-11 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 Aug 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

Gemini 4 Argon leads with 1,102

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. 1,102 Gemini 4 Argon
  2. 943 Gemini 3.8 Flash
  3. 930 GPT-6 Astra
  4. 876 Claude Opus 5
  5. 816 Gemini 3.7 Flash
  6. 790 GPT-5.5
  7. 784 GPT-6.1 Sol
  8. 753 GPT-5.4
  9. 740 Claude Fable 5.1
  10. 712 Gemini 3.5 Flash
  11. 710 Gemini 3.1 Pro
  12. 707 Gemini 3.6 Flash
  13. 692 Claude Opus 5.5
  14. 687 Claude Haiku 5.5
  15. 616 GPT-5.2
  16. 590 GPT-5.6 Sol
  17. 532 Claude Sonnet 5.5
  18. 504 GPT-6 Sol
  19. 471 Grok 4.7
  20. 457 o3
  21. 381 Claude Opus 4.8
  22. 358 Grok 4.6
  23. 336 Gemini 3 Flash
  24. 328 Claude Opus 4.6
  25. 256 Claude Sonnet 5
  26. 239 Grok 4.5
  27. 228 Claude Sonnet 4.6
  28. 206 GPT-5 mini
  29. 184 GPT-6 Luna
  30. 153 Claude Sonnet 4.5
  31. 151 GPT-5.6 Terra
  32. 124 Grok 4.20
  33. 95 GPT-5.4 Mini
  34. 40 Claude Opus 4.7
  35. 17 Grok 4.3
  36. 0 GPT-5.6 Luna
  1. 1 Gemini 4 Argon 1,102
  2. 2 Gemini 3.8 Flash 943
  3. 3 GPT-6 Astra 930
  4. 4 Claude Opus 5 876
  5. 5 Gemini 3.7 Flash 816
  6. 6 GPT-5.5 790
  7. 7 GPT-6.1 Sol 784
  8. 8 GPT-5.4 753
  9. 9 Claude Fable 5.1 740
  10. 10 Gemini 3.5 Flash 712
  11. 11 Gemini 3.1 Pro 710
  12. 12 Gemini 3.6 Flash 707
  13. 13 Claude Opus 5.5 692
  14. 14 Claude Haiku 5.5 687
  15. 15 GPT-5.2 616
  16. 16 GPT-5.6 Sol 590
  17. 17 Claude Sonnet 5.5 532
  18. 18 GPT-6 Sol 504
  19. 19 Grok 4.7 471
  20. 20 o3 457
  21. 21 Claude Opus 4.8 381
  22. 22 Grok 4.6 358
  23. 23 Gemini 3 Flash 336
  24. 24 Claude Opus 4.6 328
  25. 25 Claude Sonnet 5 256
  26. 26 Grok 4.5 239
  27. 27 Claude Sonnet 4.6 228
  28. 28 GPT-5 mini 206
  29. 29 GPT-6 Luna 184
  30. 30 Claude Sonnet 4.5 153
  31. 31 GPT-5.6 Terra 151
  32. 32 Grok 4.20 reasoning 124
  33. 33 GPT-5.4 Mini 95
  34. 34 Claude Opus 4.7 40
  35. 35 Grok 4.3 17
  36. 36 GPT-5.6 Luna 0
#ModelScoreSettingDateSource
1 Gemini 4 Argon 1,102 Sep 2026 Official leaderboard
2 Gemini 3.8 Flash 943 Sep 2026 Official leaderboard
3 GPT-6 Astra 930 Sep 2026 Official leaderboard
4 Claude Opus 5 876 Jul 2026 Official leaderboard
5 Gemini 3.7 Flash 816 Aug 2026 Official leaderboard
6 GPT-5.5 790 Apr 2026 Official leaderboard
7 GPT-6.1 Sol 784 Sep 2026 Official leaderboard
8 GPT-5.4 753 Mar 2026 Official leaderboard
9 Claude Fable 5.1 740 Sep 2026 Official leaderboard
10 Gemini 3.5 Flash 712 May 2026 Official leaderboard
11 Gemini 3.1 Pro 710 Feb 2026 Official leaderboard
12 Gemini 3.6 Flash 707 Jul 2026 Official leaderboard
13 Claude Opus 5.5 692 Sep 2026 Official leaderboard
14 Claude Haiku 5.5 687 Oct 2026 Official leaderboard
15 GPT-5.2 616 Dec 2025 Official leaderboard
16 GPT-5.6 Sol 590 Jul 2026 Official leaderboard
17 Claude Sonnet 5.5 532 Sep 2026 Official leaderboard
18 GPT-6 Sol 504 Sep 2026 Official leaderboard
19 Grok 4.7 471 Sep 2026 Official leaderboard
20 o3 457 Apr 2025 Official leaderboard
21 Claude Opus 4.8 381 May 2026 Official leaderboard
22 Grok 4.6 358 Aug 2026 Official leaderboard
23 Gemini 3 Flash 336 Dec 2025 Official leaderboard
24 Claude Opus 4.6 328 Feb 2026 Official leaderboard
25 Claude Sonnet 5 256 Jun 2026 Official leaderboard
26 Grok 4.5 239 Jul 2026 Official leaderboard
27 Claude Sonnet 4.6 228 Feb 2026 Official leaderboard
28 GPT-5 mini 206 Aug 2025 Official leaderboard
29 GPT-6 Luna 184 Sep 2026 Official leaderboard
30 Claude Sonnet 4.5 153 Sep 2025 Official leaderboard
31 GPT-5.6 Terra 151 Jul 2026 Official leaderboard
32 Grok 4.20 124 reasoning Feb 2026 Official leaderboard
33 GPT-5.4 Mini 95 Mar 2026 Official leaderboard
34 Claude Opus 4.7 40 Apr 2026 Official leaderboard
35 Grok 4.3 17 Apr 2026 Official leaderboard
36 GPT-5.6 Luna 0 Jul 2026 Official leaderboard

3 models have held number one since its release in Aug 2026

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

Anthropic Google DeepMind
Own leaderboard 3 leaders
Gemini 4 Argon 1,102

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