DiCoBench - Reasoning Difference: leaderboard
Metric: Accuracy (%) on the reasoning-difference task (traces that imply a latent physical event) of DiCoBench (765 multi-image samples at near-2K resolution): 5-option multiple choice (four candidate image pairs or instructions plus a no-visible-difference option, chance 20%), answer letter matched exactly, VLMEvalKit settings at temperature 0; higher is better. Source: arxiv.org. Saturation forecast: Around 2037. 18 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-4.1 Mini | 28 |
| 2 | Gemini 3 Pro | 28 |
| 3 | Qwen 3.5 35B A3B | 26.7 |
| 4 | GPT-4o | 25.33 |
| 5 | Qwen 2.5 VL 7B Instruct | 25.3 |
| 6 | GPT-5 | 22.7 |
| 7 | O4 Mini | 22.7 |
| 8 | Gemma 3 27B (IT) | 21.3 |
| 9 | GPT-4.1 | 21.3 |
| 10 | Gemma 3 12B (IT) | 20 |
| 11 | Qwen 2.5 VL 32B Instruct | 20 |
| 12 | Gemini 3 Flash | 14.3 |
Interactive version: theaggregate.ai/benchmark?slug=dicobench-reasoning-difference · How It Works · Data refreshed daily, snapshot 2026-09-29.