Benchmark Almanac

Game Arena Four in a Row

Gemini 4 Carbon 1,003 leads since October 2026
Not in model cards

Language models play a Connect Four style game and try to line up four of their tokens before the opponent does.

Kaggle published the leaderboard on February 19th, 2026 with 38 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
February 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
43
Models scored on its own leaderboard

Gemini 4 Carbon leads with 1,003

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.

43 models. Scroll sideways for the rest.

  1. 1,003 Gemini 4 Carbon
  2. 916 Gemini 4 Argon
  3. 819 Claude Fable 5.2
  4. 701 GPT-6 Astra
  5. 698 Gemini 3.8 Flash
  6. 686 Claude Opus 5
  7. 608 Claude Fable 5.1
  8. 605 Gemini 3.1 Pro
  9. 584 GPT-6.1 Sol
  10. 559 Gemini 3.5 Flash
  11. 552 Gemini 3.7 Flash
  12. 542 Claude Opus 5.5
  13. 504 Gemini 3.6 Flash
  14. 502 Grok 4.6
  15. 474 Grok 4.7
  16. 466 GPT-5.5
  17. 436 Claude Opus 4.8
  18. 434 GPT-5.4
  19. 419 Claude Haiku 5.5
  20. 412 Claude Sonnet 5.5
  21. 396 GPT-5.2
  22. 368 Grok 4.5
  23. 353 GPT-5.6 Sol
  24. 340 GPT-6 Sol
  25. 333 GPT-6.1 Terra
  26. 322 Claude Sonnet 5
  27. 302 Claude Opus 4.5
  28. 293 o3
  29. 290 Grok 4.20
  30. 265 Grok 4
  31. 258 Gemini 3 Flash
  32. 211 Claude Opus 4.6
  33. 204 GPT-5.4 Mini
  34. 198 Grok 4.1 Fast
  35. 196 Claude Opus 4.7
  36. 195 Grok 4.3
  37. 187 GPT-5.6 Luna
  38. 157 Claude Sonnet 4.5
  39. 133 Claude Sonnet 4.6
  40. 128 GPT-6 Luna
  41. 92 GPT-5.6 Terra
  42. 9 DeepSeek-V3.2
  43. 0 GPT-5 mini
  1. 1 Gemini 4 Carbon 1,003
  2. 2 Gemini 4 Argon 916
  3. 3 Claude Fable 5.2 819
  4. 4 GPT-6 Astra 701
  5. 5 Gemini 3.8 Flash 698
  6. 6 Claude Opus 5 686
  7. 7 Claude Fable 5.1 608
  8. 8 Gemini 3.1 Pro 605
  9. 9 GPT-6.1 Sol 584
  10. 10 Gemini 3.5 Flash 559
  11. 11 Gemini 3.7 Flash 552
  12. 12 Claude Opus 5.5 542
  13. 13 Gemini 3.6 Flash 504
  14. 14 Grok 4.6 502
  15. 15 Grok 4.7 474
  16. 16 GPT-5.5 466
  17. 17 Claude Opus 4.8 436
  18. 18 GPT-5.4 434
  19. 19 Claude Haiku 5.5 419
  20. 20 Claude Sonnet 5.5 412
  21. 21 GPT-5.2 396
  22. 22 Grok 4.5 368
  23. 23 GPT-5.6 Sol 353
  24. 24 GPT-6 Sol 340
  25. 25 GPT-6.1 Terra 333
  26. 26 Claude Sonnet 5 322
  27. 27 Claude Opus 4.5 302
  28. 28 o3 293
  29. 29 Grok 4.20 reasoning 290
  30. 30 Grok 4 265
  31. 31 Gemini 3 Flash 258
  32. 32 Claude Opus 4.6 211
  33. 33 GPT-5.4 Mini 204
  34. 34 Grok 4.1 Fast reasoning 198
  35. 35 Claude Opus 4.7 196
  36. 36 Grok 4.3 195
  37. 37 GPT-5.6 Luna 187
  38. 38 Claude Sonnet 4.5 157
  39. 39 Claude Sonnet 4.6 133
  40. 40 GPT-6 Luna 128
  41. 41 GPT-5.6 Terra 92
  42. 42 DeepSeek-V3.2 9
  43. 43 GPT-5 mini 0

Source: Official Game Arena Four in a Row leaderboard

#ModelScoreSettingDateSource
1 Gemini 4 Carbon 1,003 Oct 2026 Official leaderboard
2 Gemini 4 Argon 916 Sep 2026 Official leaderboard
3 Claude Fable 5.2 819 Nov 2026 Official leaderboard
4 GPT-6 Astra 701 Sep 2026 Official leaderboard
5 Gemini 3.8 Flash 698 Sep 2026 Official leaderboard
6 Claude Opus 5 686 Jul 2026 Official leaderboard
7 Claude Fable 5.1 608 Sep 2026 Official leaderboard
8 Gemini 3.1 Pro 605 Feb 2026 Official leaderboard
9 GPT-6.1 Sol 584 Sep 2026 Official leaderboard
10 Gemini 3.5 Flash 559 May 2026 Official leaderboard
11 Gemini 3.7 Flash 552 Aug 2026 Official leaderboard
12 Claude Opus 5.5 542 Sep 2026 Official leaderboard
13 Gemini 3.6 Flash 504 Jul 2026 Official leaderboard
14 Grok 4.6 502 Aug 2026 Official leaderboard
15 Grok 4.7 474 Sep 2026 Official leaderboard
16 GPT-5.5 466 Apr 2026 Official leaderboard
17 Claude Opus 4.8 436 May 2026 Official leaderboard
18 GPT-5.4 434 Mar 2026 Official leaderboard
19 Claude Haiku 5.5 419 Oct 2026 Official leaderboard
20 Claude Sonnet 5.5 412 Sep 2026 Official leaderboard
21 GPT-5.2 396 Dec 2025 Official leaderboard
22 Grok 4.5 368 Jul 2026 Official leaderboard
23 GPT-5.6 Sol 353 Jul 2026 Official leaderboard
24 GPT-6 Sol 340 Sep 2026 Official leaderboard
25 GPT-6.1 Terra 333 Oct 2026 Official leaderboard
26 Claude Sonnet 5 322 Jun 2026 Official leaderboard
27 Claude Opus 4.5 302 Nov 2025 Official leaderboard
28 o3 293 Apr 2025 Official leaderboard
29 Grok 4.20 290 reasoning Feb 2026 Official leaderboard
30 Grok 4 265 Jul 2025 Official leaderboard
31 Gemini 3 Flash 258 Dec 2025 Official leaderboard
32 Claude Opus 4.6 211 Feb 2026 Official leaderboard
33 GPT-5.4 Mini 204 Mar 2026 Official leaderboard
34 Grok 4.1 Fast 198 reasoning Nov 2025 Official leaderboard
35 Claude Opus 4.7 196 Apr 2026 Official leaderboard
36 Grok 4.3 195 Apr 2026 Official leaderboard
37 GPT-5.6 Luna 187 Jul 2026 Official leaderboard
38 Claude Sonnet 4.5 157 Sep 2025 Official leaderboard
39 Claude Sonnet 4.6 133 Feb 2026 Official leaderboard
40 GPT-6 Luna 128 Sep 2026 Official leaderboard
41 GPT-5.6 Terra 92 Jul 2026 Official leaderboard
42 DeepSeek-V3.2 9 Dec 2025 Official leaderboard
43 GPT-5 mini 0 Aug 2025 Official leaderboard

7 models have held number one on the official leaderboard since its release in February 2026

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

OpenAI Google DeepMind Anthropic
Own leaderboard 7 leaders
Gemini 4 Carbon 1,003

Source: Official Game Arena Four in a Row 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
0 250 500 750 1,000 1,250 2025 2026 2027 0 250 500 750 1,000 1,250 2025 2026 2027

Source: Official Game Arena Four in a Row 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%