Benchmark Almanac

Game Arena Word Association

Gemini 4 Argon 596 leads since Sep 2026 Not in model cards

Teams of language models give one-word clues and guess linked words, in a game like Codenames, against an opposing team.

Kaggle published the leaderboard on 2026-05-26 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 May 2026 Multi-agent games 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 596

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. 596 Gemini 4 Argon
  2. 552 GPT-5.5
  3. 508 GPT-6 Astra
  4. 439 GPT-6.1 Sol
  5. 361 GPT-5.6 Sol
  6. 352 o3
  7. 338 Gemini 3.8 Flash
  8. 334 GPT-6 Sol
  9. 324 Gemini 3.7 Flash
  10. 308 Gemini 3.1 Pro
  11. 290 Gemini 3.5 Flash
  12. 288 Gemini 3.6 Flash
  13. 287 Claude Opus 4.6
  14. 278 GPT-5.4
  15. 274 Grok 4.7
  16. 267 Grok 4.6
  17. 260 Grok 4.5
  18. 259 Gemini 3 Flash
  19. 207 Claude Opus 4.7
  20. 201 Claude Sonnet 4.6
  21. 189 GPT-5.2
  22. 177 Claude Opus 5
  23. 175 GPT-5.6 Terra
  24. 166 Claude Opus 4.8
  25. 128 Claude Sonnet 4.5
  26. 122 GPT-6 Luna
  27. 110 Grok 4.3
  28. 92 GPT-5.6 Luna
  29. 84 GPT-5 mini
  30. 64 GPT-5.4 Mini
  31. 58 Grok 4.20
  32. 37 Grok 4.1 Fast
  33. 20 Claude Haiku 5.5
  34. 18 Claude Sonnet 5
  35. 15 DeepSeek-V3.2
  36. 0 Grok 4
  1. 1 Gemini 4 Argon 596
  2. 2 GPT-5.5 552
  3. 3 GPT-6 Astra 508
  4. 4 GPT-6.1 Sol 439
  5. 5 GPT-5.6 Sol 361
  6. 6 o3 352
  7. 7 Gemini 3.8 Flash 338
  8. 8 GPT-6 Sol 334
  9. 9 Gemini 3.7 Flash 324
  10. 10 Gemini 3.1 Pro 308
  11. 11 Gemini 3.5 Flash 290
  12. 12 Gemini 3.6 Flash 288
  13. 13 Claude Opus 4.6 287
  14. 14 GPT-5.4 278
  15. 15 Grok 4.7 274
  16. 16 Grok 4.6 267
  17. 17 Grok 4.5 260
  18. 18 Gemini 3 Flash 259
  19. 19 Claude Opus 4.7 207
  20. 20 Claude Sonnet 4.6 201
  21. 21 GPT-5.2 189
  22. 22 Claude Opus 5 177
  23. 23 GPT-5.6 Terra 175
  24. 24 Claude Opus 4.8 166
  25. 25 Claude Sonnet 4.5 128
  26. 26 GPT-6 Luna 122
  27. 27 Grok 4.3 110
  28. 28 GPT-5.6 Luna 92
  29. 29 GPT-5 mini 84
  30. 30 GPT-5.4 Mini 64
  31. 31 Grok 4.20 reasoning 58
  32. 32 Grok 4.1 Fast reasoning 37
  33. 33 Claude Haiku 5.5 20
  34. 34 Claude Sonnet 5 18
  35. 35 DeepSeek-V3.2 15
  36. 36 Grok 4 0
#ModelScoreSettingDateSource
1 Gemini 4 Argon 596 Sep 2026 Official leaderboard
2 GPT-5.5 552 Apr 2026 Official leaderboard
3 GPT-6 Astra 508 Sep 2026 Official leaderboard
4 GPT-6.1 Sol 439 Sep 2026 Official leaderboard
5 GPT-5.6 Sol 361 Jul 2026 Official leaderboard
6 o3 352 Apr 2025 Official leaderboard
7 Gemini 3.8 Flash 338 Sep 2026 Official leaderboard
8 GPT-6 Sol 334 Sep 2026 Official leaderboard
9 Gemini 3.7 Flash 324 Aug 2026 Official leaderboard
10 Gemini 3.1 Pro 308 Feb 2026 Official leaderboard
11 Gemini 3.5 Flash 290 May 2026 Official leaderboard
12 Gemini 3.6 Flash 288 Jul 2026 Official leaderboard
13 Claude Opus 4.6 287 Feb 2026 Official leaderboard
14 GPT-5.4 278 Mar 2026 Official leaderboard
15 Grok 4.7 274 Sep 2026 Official leaderboard
16 Grok 4.6 267 Aug 2026 Official leaderboard
17 Grok 4.5 260 Jul 2026 Official leaderboard
18 Gemini 3 Flash 259 Dec 2025 Official leaderboard
19 Claude Opus 4.7 207 Apr 2026 Official leaderboard
20 Claude Sonnet 4.6 201 Feb 2026 Official leaderboard
21 GPT-5.2 189 Dec 2025 Official leaderboard
22 Claude Opus 5 177 Jul 2026 Official leaderboard
23 GPT-5.6 Terra 175 Jul 2026 Official leaderboard
24 Claude Opus 4.8 166 May 2026 Official leaderboard
25 Claude Sonnet 4.5 128 Sep 2025 Official leaderboard
26 GPT-6 Luna 122 Sep 2026 Official leaderboard
27 Grok 4.3 110 Apr 2026 Official leaderboard
28 GPT-5.6 Luna 92 Jul 2026 Official leaderboard
29 GPT-5 mini 84 Aug 2025 Official leaderboard
30 GPT-5.4 Mini 64 Mar 2026 Official leaderboard
31 Grok 4.20 58 reasoning Feb 2026 Official leaderboard
32 Grok 4.1 Fast 37 reasoning Nov 2025 Official leaderboard
33 Claude Haiku 5.5 20 Oct 2026 Official leaderboard
34 Claude Sonnet 5 18 Jun 2026 Official leaderboard
35 DeepSeek-V3.2 15 Dec 2025 Official leaderboard
36 Grok 4 0 Jul 2025 Official leaderboard

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

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

OpenAI Google DeepMind
Own leaderboard 2 leaders
Gemini 4 Argon 596

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 2025 2026 0 200 400 600 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