MM-JudgeBias: 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, and should lower its score when evidence is removed or mismatched and keep it when the change preserves meaning. Mean reliability from 0 to 1 over nine bias types; higher is better.
Metric: Mean judge-reliability score (0 to 1) over nine bias types 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): Bias-Deviation (normalized score drop when the image or query is removed or mismatched) for five types and Bias-Conformity (score invariance under meaning-preserving perturbations) for four; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 26 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Setting | Score | Overall rank |
|---|---|---|---|---|
| 1 | Gemini 2.5 Pro | 0.869 | #276 | |
| 2 | Gemini 3 Pro | High | 0.869 | #145 |
| 3 | Claude Opus 4.5 | Thinking | 0.858 | #160 |
| 4 | Claude Sonnet 4.5 | Thinking | 0.816 | #261 |
| 5 | Gemini 2.5 Flash | 0.761 | #463 | |
| 6 | Claude Haiku 4.5 | Thinking | 0.757 | #583 |
| 7 | Gemini 2.5 Flash Lite | Thinking | 0.747 | #964 |
| 8 | Qwen 3 VL 8B Instruct | 0.738 | #1029 | |
| 9 | Qwen 3 VL 30B A3B (Thinking) | 0.713 | #719 | |
| 10 | Qwen 3 VL 30B A3B Instruct | 0.678 | #895 | |
| 11 | O3 | High | 0.675 | #237 |
| 12 | Qwen 3 VL 8B (Thinking) | 0.655 | #910 | |
| 13 | O4 Mini | High | 0.654 | #368 |
| 14 | Gemini 2.0 Flash Lite | 0.647 | #1050 | |
| 15 | LLaVA-Critic-7B | 0.647 |
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 · How It Works · Data refreshed daily, snapshot 2026-11-21.