MM-JudgeBias - Text Dominance: leaderboard

MM-JudgeBias (2026) tests multimodal LLM judges for compositional bias: the judge scores image, query and response triplets from 1 to 10 before and after a controlled perturbation. Bias-Deviation from 0 to 1, the normalized score drop when the image is replaced with a black image; higher is better.

Metric: Bias-Deviation (0 to 1) on MM-JudgeBias (about 200 image-query-response triplets per bias type, 1,804 in all, each judged on a 1-10 scale before and after a controlled perturbation; mean of three runs): mean drop of the judge's score, normalized by the largest possible drop, when the image is replaced with a black image; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 26 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1Gemini 3 ProHigh0.912#145
2Gemini 2.5 Pro0.751#276
3Claude Opus 4.5Thinking0.68#160
4Claude Sonnet 4.5Thinking0.556#261
5Gemini 2.5 Flash0.423#463
6Gemini 2.5 Flash LiteThinking0.391#964
7Qwen 3 VL 8B Instruct0.336#1029
8Qwen 3 VL 30B A3B (Thinking)0.293#719
9Claude Haiku 4.5Thinking0.291#583
10O3High0.276#237
11LLaVA-Critic-7B0.238
12Qwen 3 VL 30B A3B Instruct0.237#895
13GPT-5.1High0.192#240
14Qwen 3 VL 8B (Thinking)0.177#910
15Prometheus-Vision-7B0.167

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

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