ChessImageBench: leaderboard
ChessImageBench — reading chess positions from board images (FEN transcription and piece identification); percent of fully correct boards.
Metric: All Correct (%). Source: github.com. Saturation forecast: Around September 2028. 5 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) (Low) | 20.8 | #58 (Gemini 3.1 Pro (Preview)) |
| 2 | Gemini 2.5 Pro | 7.2 | #152 |
| 3 | GPT-5.4 (Low) | 6.7 | #79 (GPT-5.4) |
| 4 | GPT-5 | 4.8 | #98 |
| 5 | Grok 4 | 4.3 | #181 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=chessimagebench · How It Works · Data refreshed daily, snapshot 2026-10-20.