Kaggle Game Arena Chess (Openings): leaderboard
LLM chess performance in the opening phase, measuring knowledge of opening theory and early positional decisions in tournament play.
Metric: Elo Rating. Source: www.kaggle.com. Saturation forecast: Around April 2027. 11 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Flash (Preview) | 1263 | #82 |
| 2 | Gemini 3 Pro (Preview) | 1253 | #66 |
| 3 | O3 | 934 | #124 |
| 4 | GPT-5.2 | 889 | #110 |
| 5 | Grok 4 | 739 | #181 |
| 6 | Grok 4.1 Fast (Reasoning) | 635 | #217 (Grok 4.1 Fast) |
| 7 | GPT-5 Mini | 484 | #186 |
| 8 | Claude Opus 4.5 | 252 | #83 |
| 9 | Claude Sonnet 4.5 | 228 | #142 |
| 10 | Claude Haiku 4.5 | 102 | #280 |
| 11 | DeepSeek V3.2 | 0 | #205 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=kaggle-game-arena-chess-openings · How It Works · Data refreshed daily, snapshot 2026-10-20.