Kaggle Game Arena (Unified): leaderboard
Kaggle Game Arena's unified leaderboard: one rating per model pooled from its matches across the Game Arena games, which include text chess, Four in a Row, heads-up poker and Werewolf; each model covered 5 to 16 games on 2026-09-13. Higher is better on an unbounded scale where the lowest-rated model sits at 0, not comparable to the per-game Elo boards.
Metric: Unified Score. Source: www.kaggle.com. Saturation forecast: Around November 2027. 39 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 4 Argon | 535.09 | #9 |
| 2 | GPT-6 Astra | 490.98 | #6 |
| 3 | Claude Opus 5 | 436.19 | #16 |
| 4 | GPT-6.1 Sol | 436.07 | #7 |
| 5 | Gemini 3.8 Flash | 428.51 | #30 |
| 6 | Claude Fable 5.1 | 418.49 | #10 |
| 7 | GPT-5.5 | 407.31 | #33 |
| 8 | Claude Opus 5.5 | 398.04 | #3 |
| 9 | Gemini 3.1 Pro (Preview) | 381.16 | #64 |
| 10 | Gemini 3.7 Flash | 368.15 | #24 |
| 11 | GPT-5.6 Sol | 352.39 | #22 |
| 12 | GPT-5.4 | 351.53 | #83 |
| 13 | Grok 4.7 | 335.17 | #26 |
| 14 | GPT-6 Sol | 330.69 | #19 |
| 15 | Gemini 3.6 Flash | 324.69 | #57 |
Interactive version: theaggregate.ai/benchmark?slug=kaggle-game-arena-unified · How It Works · Data refreshed daily, snapshot 2026-11-05.