DistortBench - Spatial: leaderboard

DistortBench (2026) shows a vision-language model a distorted photo without its original and asks it to name the distortion type and severity among four options (chance 25 percent). Joint accuracy on the eight spatial distortions (jitter, non-eccentricity, pixelate, quantization, color block, sharpen, two rotations), in percent; higher is better.

Metric: Joint accuracy (%) on the eight spatial distortions (jitter, non-eccentricity, pixelate, quantization, color block, sharpen, two rotations), a category of DistortBench's 13,500 four-choice questions (27 distortion types at five KADID-10k calibrated severity levels on 100 KADIS-700k photos, no reference image), the answer naming both type and severity, answered-only accuracy (unparseable responses, at most 3.4 percent, leave the denominator), chance 25; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 14 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1Qwen 3.5 27BNon-reasoning64.8#380
2Qwen 3.5 35B A3B60.1#494
3Qwen 3.5 4BNon-reasoning58.1#1270
4Qwen 3.5 9B57.8#838
5Kimi-VL-A3B-Instruct53#2272
6InternVL3.5-38B52.5#1026
7GPT-5.449.8#138
8Gemma 3 27B46.8#1672
9Gemma 3 12B46.3#2091
10Kimi-VL-A3B (Thinking)44.5
11InternVL3.5-8B41.5#1618
12InternVL3.5-14B37#1519
13Qwen 2 VL 7B35.9#2300
14Gemma 3 4B33.4#5347

No result here: #2 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=distortbench-spatial · How It Works · Data refreshed daily, snapshot 2026-11-22.