DistortBench - Blur: 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 three blur types (Gaussian, lens, motion), in percent; higher is better.

Metric: Joint accuracy (%) on the three blur types (Gaussian, lens, motion), 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 35B A3B68.3#492
2Qwen 3.5 4BNon-reasoning65.5#1241
3InternVL3.5-38B64.6#1021
4GPT-5.463.9#135
5Kimi-VL-A3B-Instruct63.7#2253
6Qwen 3.5 27B63#375
7Qwen 3.5 9B59.4#828
8Gemma 3 27B59#1663
9Gemma 3 12B58.6#2059
10InternVL3.5-8B58.3#1604
11InternVL3.5-14B54.3#1504
12Kimi-VL-A3B (Thinking)51.8
13Gemma 3 4B45.8#5339
14Qwen 2 VL 7B44.1#2275

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=distortbench-blur · How It Works · Data refreshed daily, snapshot 2026-11-21.