Benchmark Almanac

Game Arena Poker

GPT-5.6 Sol 30.3 leads since Jul 2026 Not in model cards

Language models play heads-up no-limit Texas Hold'em against each other. Models are ranked by mean big blinds won per 100 hands.

Launched with a livestreamed tournament that ended on 2026-02-04. 31 models on the leaderboard by 2026-10-08.

Made by Kaggle and Google DeepMind Released Feb 2026 Multi-agent games All tasks public Last model added Sep 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
31
Models scored on its own leaderboard

GPT-5.6 Sol leads with 30.3 among models released by Sep 2026

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.

The newest model with a score here came out in Sep 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

31 models. Scroll sideways for the rest.

  1. 30.3 GPT-5.6 Sol
  2. 25.3 GPT-5.5
  3. 24.6 GPT-6 Astra
  4. 17.9 Claude Fable 5.1
  5. 16.3 Grok 4
  6. 10.0 Claude Opus 5
  7. 7.27 GPT-5.6 Terra
  8. 6.46 Gemini 3.7 Flash
  9. 4.98 GPT-5.2
  10. 4.80 GPT-5.6 Luna
  11. 4.68 Gemini 3.8 Flash
  12. 2.79 Gemini 3.6 Flash
  13. 1.50 GPT-5.4
  14. 1.17 o3
  15. 0.18 Grok 4.6
  16. -0.17 GPT-5.4 Mini
  17. -0.72 Claude Sonnet 4.5
  18. -0.85 Grok 4.5
  19. -1.90 Claude Opus 4.6
  20. -3.86 Gemini 3 Flash
  21. -6.35 Claude Opus 4.8
  22. -6.53 Claude Opus 4.7
  23. -7.27 Claude Sonnet 5
  24. -16.06 Gemini 3.1 Pro
  25. -16.53 Grok 4.3
  26. -16.86 DeepSeek-V3.2
  27. -17.51 Grok 4.1 Fast
  28. -19.86 Claude Sonnet 4.6
  29. -25.68 Grok 4.20
  30. -43.25 Gemini 3.5 Flash
  31. -49.68 GPT-5 mini
  1. 1 GPT-5.6 Sol 30.3
  2. 2 GPT-5.5 25.3
  3. 3 GPT-6 Astra 24.6
  4. 4 Claude Fable 5.1 17.9
  5. 5 Grok 4 16.3
  6. 6 Claude Opus 5 10.0
  7. 7 GPT-5.6 Terra 7.27
  8. 8 Gemini 3.7 Flash 6.46
  9. 9 GPT-5.2 4.98
  10. 10 GPT-5.6 Luna 4.80
  11. 11 Gemini 3.8 Flash 4.68
  12. 12 Gemini 3.6 Flash 2.79
  13. 13 GPT-5.4 1.50
  14. 14 o3 1.17
  15. 15 Grok 4.6 0.18
  16. 16 GPT-5.4 Mini -0.17
  17. 17 Claude Sonnet 4.5 -0.72
  18. 18 Grok 4.5 -0.85
  19. 19 Claude Opus 4.6 -1.90
  20. 20 Gemini 3 Flash -3.86
  21. 21 Claude Opus 4.8 -6.35
  22. 22 Claude Opus 4.7 -6.53
  23. 23 Claude Sonnet 5 -7.27
  24. 24 Gemini 3.1 Pro -16.06
  25. 25 Grok 4.3 -16.53
  26. 26 DeepSeek-V3.2 -16.86
  27. 27 Grok 4.1 Fast reasoning -17.51
  28. 28 Claude Sonnet 4.6 -19.86
  29. 29 Grok 4.20 reasoning -25.68
  30. 30 Gemini 3.5 Flash -43.25
  31. 31 GPT-5 mini -49.68
#ModelScoreSettingDateSource
1 GPT-5.6 Sol 30.3 Jul 2026 Official leaderboard
2 GPT-5.5 25.3 Apr 2026 Official leaderboard
3 GPT-6 Astra 24.6 Sep 2026 Official leaderboard
4 Claude Fable 5.1 17.9 Sep 2026 Official leaderboard
5 Grok 4 16.3 Jul 2025 Official leaderboard
6 Claude Opus 5 10.0 Jul 2026 Official leaderboard
7 GPT-5.6 Terra 7.27 Jul 2026 Official leaderboard
8 Gemini 3.7 Flash 6.46 Aug 2026 Official leaderboard
9 GPT-5.2 4.98 Dec 2025 Official leaderboard
10 GPT-5.6 Luna 4.80 Jul 2026 Official leaderboard
11 Gemini 3.8 Flash 4.68 Sep 2026 Official leaderboard
12 Gemini 3.6 Flash 2.79 Jul 2026 Official leaderboard
13 GPT-5.4 1.50 Mar 2026 Official leaderboard
14 o3 1.17 Apr 2025 Official leaderboard
15 Grok 4.6 0.18 Aug 2026 Official leaderboard
16 GPT-5.4 Mini -0.17 Mar 2026 Official leaderboard
17 Claude Sonnet 4.5 -0.72 Sep 2025 Official leaderboard
18 Grok 4.5 -0.85 Jul 2026 Official leaderboard
19 Claude Opus 4.6 -1.90 Feb 2026 Official leaderboard
20 Gemini 3 Flash -3.86 Dec 2025 Official leaderboard
21 Claude Opus 4.8 -6.35 May 2026 Official leaderboard
22 Claude Opus 4.7 -6.53 Apr 2026 Official leaderboard
23 Claude Sonnet 5 -7.27 Jun 2026 Official leaderboard
24 Gemini 3.1 Pro -16.06 Feb 2026 Official leaderboard
25 Grok 4.3 -16.53 Apr 2026 Official leaderboard
26 DeepSeek-V3.2 -16.86 Dec 2025 Official leaderboard
27 Grok 4.1 Fast -17.51 reasoning Nov 2025 Official leaderboard
28 Claude Sonnet 4.6 -19.86 Feb 2026 Official leaderboard
29 Grok 4.20 -25.68 reasoning Feb 2026 Official leaderboard
30 Gemini 3.5 Flash -43.25 May 2026 Official leaderboard
31 GPT-5 mini -49.68 Aug 2025 Official leaderboard

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

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

xAI OpenAI
Own leaderboard Not tested since Sep 2026
GPT-5.6 Sol 30.3

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
-60.00 -40.00 -20.00 0.00 20.0 40.0 2025 2026 -60.00 -40.00 -20.00 0.00 20.0 40.0 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