OddGridBench - Three Attributes: leaderboard
Metric: Accuracy (%) on the 200 images in which the odd item differs in three attributes at once; OddGridBench synthetic grid images (5 to 9 rows and columns of identical icons with one odd item); the model names the odd item's row and column, scored by exact match; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 19 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3 VL 32B Instruct | 89.5 | #276 |
| 2 | Gemini 2.5 Pro | 76 | #145 |
| 3 | Qwen 3 VL 4B Instruct | 70.5 | #506 |
| 4 | Qwen 3 VL 8B Instruct | 67 | #401 |
| 5 | Gemini 2.5 Flash | 54.5 | #237 |
| 6 | Qwen 3 VL 30B A3B Instruct | 52.5 | #365 |
| 7 | Gemini 2.0 Flash | 43.5 | #331 |
| 8 | GPT-5 (Low) | 35.5 | #91 (GPT-5) |
| 9 | Qwen 2.5 VL 72B Instruct | 17.5 | #364 |
| 10 | Qwen 2.5 VL 7B Instruct | 2.5 | #643 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=oddgridbench-three-attributes · How It Works · Data refreshed daily, snapshot 2026-10-11.