VRIQ (Natural) - Figure Rotation: leaderboard
Metric: Accuracy (%) on the 100 natural-domain VRIQ figure rotation puzzles (the same reasoning categories over real-world objects and scenes, seeded by annotators and generated with GPT-5.2-Thinking or sourced, human validated; three options); higher is better. Source: arxiv.org. Saturation forecast: Around 2032. 14 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3 VL 32B (Thinking) | 49 | #287 (Qwen 3 VL 32B) |
| 2 | GPT-4o | 40 | #333 |
| 3 | GPT-4o Mini | 40 | #588 |
| 4 | GPT-5 Mini | 40 | #176 |
| 5 | GPT-5.1 | 40 | #131 |
| 6 | Qwen 2.5 VL 7B Instruct | 35 | #643 |
| 7 | Gemini 2.5 Pro | 33 | #145 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=vriq-natural-figure-rotation · How It Works · Data refreshed daily, snapshot 2026-10-11.