Benchmark Almanac

Game Arena Word Art

Gemini 4 Argon 2,473 leads since Sep 2026 Not in model cards

Language models take turns drawing a word as ASCII art and guessing what the other model drew.

Kaggle published the leaderboard on 2026-08-18 with 35 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 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
35
Models scored on its own leaderboard

Gemini 4 Argon leads with 2,473

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.

35 models. Scroll sideways for the rest.

  1. 2,473 Gemini 4 Argon
  2. 2,169 GPT-6 Astra
  3. 1,968 GPT-6.1 Sol
  4. 1,570 Claude Opus 5.5
  5. 1,405 GPT-6 Sol
  6. 1,389 Gemini 3.8 Flash
  7. 1,100 Gemini 3.1 Pro
  8. 1,074 GPT-5.6 Sol
  9. 1,059 Gemini 3.7 Flash
  10. 1,021 GPT-5.5
  11. 797 Claude Opus 5
  12. 678 Gemini 3 Flash
  13. 642 Gemini 3.5 Flash
  14. 620 Gemini 3.6 Flash
  15. 604 Claude Sonnet 5.5
  16. 603 GPT-5.4
  17. 492 Grok 4.7
  18. 451 Claude Opus 4.6
  19. 367 Claude Opus 4.8
  20. 328 Grok 4.6
  21. 325 Claude Opus 4.7
  22. 295 GPT-5.6 Terra
  23. 276 o3
  24. 259 GPT-5.6 Luna
  25. 254 GPT-6 Luna
  26. 248 GPT-5.2
  27. 219 Claude Sonnet 4.6
  28. 218 Claude Sonnet 4.5
  29. 178 Grok 4.5
  30. 146 Claude Sonnet 5
  31. 126 GPT-5.4 Mini
  32. 102 Claude Haiku 5.5
  33. 67 Grok 4.20
  34. 2 Grok 4.3
  35. 0 GPT-5 mini
  1. 1 Gemini 4 Argon 2,473
  2. 2 GPT-6 Astra 2,169
  3. 3 GPT-6.1 Sol 1,968
  4. 4 Claude Opus 5.5 1,570
  5. 5 GPT-6 Sol 1,405
  6. 6 Gemini 3.8 Flash 1,389
  7. 7 Gemini 3.1 Pro 1,100
  8. 8 GPT-5.6 Sol 1,074
  9. 9 Gemini 3.7 Flash 1,059
  10. 10 GPT-5.5 1,021
  11. 11 Claude Opus 5 797
  12. 12 Gemini 3 Flash 678
  13. 13 Gemini 3.5 Flash 642
  14. 14 Gemini 3.6 Flash 620
  15. 15 Claude Sonnet 5.5 604
  16. 16 GPT-5.4 603
  17. 17 Grok 4.7 492
  18. 18 Claude Opus 4.6 451
  19. 19 Claude Opus 4.8 367
  20. 20 Grok 4.6 328
  21. 21 Claude Opus 4.7 325
  22. 22 GPT-5.6 Terra 295
  23. 23 o3 276
  24. 24 GPT-5.6 Luna 259
  25. 25 GPT-6 Luna 254
  26. 26 GPT-5.2 248
  27. 27 Claude Sonnet 4.6 219
  28. 28 Claude Sonnet 4.5 218
  29. 29 Grok 4.5 178
  30. 30 Claude Sonnet 5 146
  31. 31 GPT-5.4 Mini 126
  32. 32 Claude Haiku 5.5 102
  33. 33 Grok 4.20 reasoning 67
  34. 34 Grok 4.3 2
  35. 35 GPT-5 mini 0
#ModelScoreSettingDateSource
1 Gemini 4 Argon 2,473 Sep 2026 Official leaderboard
2 GPT-6 Astra 2,169 Sep 2026 Official leaderboard
3 GPT-6.1 Sol 1,968 Sep 2026 Official leaderboard
4 Claude Opus 5.5 1,570 Sep 2026 Official leaderboard
5 GPT-6 Sol 1,405 Sep 2026 Official leaderboard
6 Gemini 3.8 Flash 1,389 Sep 2026 Official leaderboard
7 Gemini 3.1 Pro 1,100 Feb 2026 Official leaderboard
8 GPT-5.6 Sol 1,074 Jul 2026 Official leaderboard
9 Gemini 3.7 Flash 1,059 Aug 2026 Official leaderboard
10 GPT-5.5 1,021 Apr 2026 Official leaderboard
11 Claude Opus 5 797 Jul 2026 Official leaderboard
12 Gemini 3 Flash 678 Dec 2025 Official leaderboard
13 Gemini 3.5 Flash 642 May 2026 Official leaderboard
14 Gemini 3.6 Flash 620 Jul 2026 Official leaderboard
15 Claude Sonnet 5.5 604 Sep 2026 Official leaderboard
16 GPT-5.4 603 Mar 2026 Official leaderboard
17 Grok 4.7 492 Sep 2026 Official leaderboard
18 Claude Opus 4.6 451 Feb 2026 Official leaderboard
19 Claude Opus 4.8 367 May 2026 Official leaderboard
20 Grok 4.6 328 Aug 2026 Official leaderboard
21 Claude Opus 4.7 325 Apr 2026 Official leaderboard
22 GPT-5.6 Terra 295 Jul 2026 Official leaderboard
23 o3 276 Apr 2025 Official leaderboard
24 GPT-5.6 Luna 259 Jul 2026 Official leaderboard
25 GPT-6 Luna 254 Sep 2026 Official leaderboard
26 GPT-5.2 248 Dec 2025 Official leaderboard
27 Claude Sonnet 4.6 219 Feb 2026 Official leaderboard
28 Claude Sonnet 4.5 218 Sep 2025 Official leaderboard
29 Grok 4.5 178 Jul 2026 Official leaderboard
30 Claude Sonnet 5 146 Jun 2026 Official leaderboard
31 GPT-5.4 Mini 126 Mar 2026 Official leaderboard
32 Claude Haiku 5.5 102 Oct 2026 Official leaderboard
33 Grok 4.20 67 reasoning Feb 2026 Official leaderboard
34 Grok 4.3 2 Apr 2026 Official leaderboard
35 GPT-5 mini 0 Aug 2025 Official leaderboard

4 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.

Google DeepMind OpenAI
Own leaderboard 4 leaders
Gemini 4 Argon 2,473

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 2,500 2025 2026 0 500 1,000 1,500 2,000 2,500 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