DiCoBench - Spatial Difference: leaderboard

Metric: Accuracy (%) on the spatial-difference task (an object moved within an aligned scene) 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 December 2026. 18 models tracked.

Top models

#ModelScore
1Gemini 3 Pro82
2Qwen 3.5 35B A3B64
3GPT-4.1 Mini54
4O4 Mini54
5GPT-548
6Gemini 3 Flash40
7Qwen 2.5 VL 32B Instruct38
8Qwen 2.5 VL 7B Instruct34
9Gemma 3 12B (IT)30
10GPT-4o26
11Gemma 3 27B (IT)24
12GPT-4.124

Interactive version: theaggregate.ai/benchmark?slug=dicobench-spatial-difference · How It Works · Data refreshed daily, snapshot 2026-09-29.