MMErroR - Error Type Classification - Biology & Healthcare: leaderboard
MMErroR (Shi et al., ACL 2026) error type classification on its Biology & Healthcare samples: naming which of four error types (visual perception, knowledge deployment, question comprehension, reasoning) the single injected error in an image-grounded reasoning chain belongs to. Accuracy (%); higher is better.
Metric: Error Type Classification Accuracy (%). Source: arxiv.org. Saturation forecast: Not forecast. 12 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen-VL-Max | 66.39 | #1226 |
| 2 | Qwen 3 VL 32B (Thinking) | 65.83 | #557 |
| 3 | Gemini 3 Pro (Preview) | 64.43 | #111 |
| 4 | GLM-4.6V | 63.87 | #698 |
| 5 | Grok 4 | 61.34 | #306 |
| 6 | GPT-5.2 (xHigh) | 60.5 | #172 (GPT-5.2) |
| 7 | Seed 2.0 Pro | 59.94 | #209 |
| 8 | Kimi K2.5 | 58.82 | #252 |
| 9 | Qwen 3 VL 32B Instruct | 58.54 | #582 |
| 10 | Claude Opus 4.5 | 57.7 | #152 |
| 11 | Llama 4 Maverick Instruct | 50.14 | #1125 |
| 12 | Llama 4 Scout Instruct | 49.02 | #1532 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=mmerror-error-type-classification-biology-healthcare · How It Works · Data refreshed daily, snapshot 2026-11-12.