VRIQ (Abstract) - Odd One Out: leaderboard
Metric: Accuracy (%) on the 200 abstract-domain VRIQ odd-one-out puzzles (four-option IQ-style multiple choice over geometric primitives, adapted from public civil-service and aptitude exams or newly designed, solved by two annotators); higher is better. Source: arxiv.org. Saturation forecast: Around 2032. 14 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.1 | 33.5 | #131 |
| 2 | Qwen 3 VL 32B (Thinking) | 33.5 | #287 (Qwen 3 VL 32B) |
| 3 | GPT-4o | 30.5 | #333 |
| 4 | GPT-4o Mini | 28.5 | #588 |
| 5 | Gemini 2.5 Pro | 27.5 | #145 |
| 6 | Qwen 2.5 VL 7B Instruct | 26.5 | #643 |
| 7 | GPT-5 Mini | 26 | #176 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=vriq-abstract-odd-one-out · How It Works · Data refreshed daily, snapshot 2026-10-11.