DiffCap-Bench - Spatial: leaderboard
DiffCap-Bench (2026) image difference captioning restricted to spatial rearrangements of objects. Macro-averaged F1* in percent, combining the share of true differences described and the share of described differences that are correct, as judged by Gemini 2.5 Pro; higher is better.
Metric: Macro-averaged F1* (%) on spatial rearrangements of objects: harmonic mean of recall* (share of key differences described) and precision* (share of described differences that are correct) on the 1,075 image pairs of DiffCap-Bench (6,713 human-verified difference items), one fixed captioning prompt, open models at temperature 0.1 and proprietary models at their defaults, a Gemini 2.5 Pro judge matching each described difference against the key difference list; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 14 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Setting | Score | Overall rank |
|---|---|---|---|---|
| 1 | Gemini 3 Pro | 83.5 | #145 | |
| 2 | GPT-5.2 | 82.7 | #185 | |
| 3 | Qwen 3 VL Plus | 78.6 | #512 | |
| 4 | Kimi K2.5 | Thinking | 78 | #266 |
| 5 | Seed1.6-VL | Thinking | 75.6 | #385 |
| 6 | Qwen 3 VL 32B (Thinking) | 75 | #573 | |
| 7 | Step3 VL 10B | 73.9 | #1080 | |
| 8 | MiMo-VL-7B-RL-2508 | 73.9 | #823 | |
| 9 | Grok 4.20 | 73.8 | #301 | |
| 10 | Qwen 3 VL 32B Instruct | 71.5 | #606 | |
| 11 | Qwen 3 VL 8B (Thinking) | 71.1 | #910 | |
| 12 | Qwen 3 VL 8B Instruct | 67 | #1029 | |
| 13 | InternVL3.5-38B | 63.1 | #1021 | |
| 14 | InternVL3.5-8B | 63 | #1604 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=diffcap-bench-spatial · How It Works · Data refreshed daily, snapshot 2026-11-21.