VisReason (Vision-Centric) - Sudoku Solving: leaderboard
Metric: Accuracy (%) on the Sudoku Solving category (structural reasoning: solve visually presented Sudoku variants) of VisReason, answers scored by format (regular-expression match, GPT-5-mini judge for open-ended answers, IoU above 0.5 for boxes), zero-shot with chain-of-thought prompt templates, one per answer format; rows not shaded gray in the paper use explicit reasoning (GPT-5 and Gemini 3 Pro at low reasoning effort); higher is better. Source: arxiv.org. Saturation forecast: Around May 2027. 22 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.2 (Thinking) | 38.2 |
| 2 | Gemini 3 Pro (Preview) (Low) | 30 |
| 3 | GPT-5 (Low) | 22.5 |
| 4 | O4 Mini | 15.5 |
| 5 | GPT-5 Mini (Low) | 15 |
| 6 | GPT-5 Mini (Medium) | 12.5 |
| 7 | Qwen 3 VL 8B Instruct | 9.2 |
| 8 | GPT-5 Nano | 9 |
| 9 | InternVL3-8B | 8.5 |
| 10 | Qwen 3 VL 30B A3B (Thinking) | 8.5 |
| 11 | GPT-5 Mini (High) | 8 |
| 12 | Qwen 2.5 VL 32B Instruct | 8 |
| 13 | GPT-4o | 7.5 |
| 14 | Qwen 3 VL 32B (Thinking) | 7 |
| 15 | Gemini 2.5 Flash | 6 |
Interactive version: theaggregate.ai/benchmark?slug=visreason-vision-centric-sudoku-solving · How It Works · Data refreshed daily, snapshot 2026-10-07.