MMErroR - Error Type Classification - Chemistry & Materials: leaderboard
Metric: Error Type Classification Accuracy (%). Source: arxiv.org. Saturation forecast: Around December 2026. 12 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Pro (Preview) | 69.81 |
| 2 | GPT-5.2 (xHigh) | 62.26 |
| 3 | Claude Opus 4.5 | 61.64 |
| 4 | Seed 2.0 Pro | 61.01 |
| 5 | Grok 4 | 59.75 |
| 6 | Qwen 3 VL 32B (Thinking) | 52.2 |
| 7 | Kimi K2.5 | 51.57 |
| 8 | Llama 4 Scout Instruct | 39.62 |
| 9 | Llama 4 Maverick Instruct | 39.62 |
| 10 | GLM-4.6V | 37.11 |
| 11 | Qwen 3 VL 32B Instruct | 28.93 |
Interactive version: theaggregate.ai/benchmark?slug=mmerror-error-type-classification-chemistry-materials · How It Works · Data refreshed daily, snapshot 2026-09-25.