MMErroR - Error Type Classification - Mathematics & Logic: leaderboard
Metric: Error Type Classification Accuracy (%). Source: arxiv.org. Saturation forecast: Around May 2027. 12 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Pro (Preview) | 66.37 |
| 2 | Seed 2.0 Pro | 65.47 |
| 3 | GPT-5.2 (xHigh) | 64.56 |
| 4 | Kimi K2.5 | 63.66 |
| 5 | Claude Opus 4.5 | 62.76 |
| 6 | Grok 4 | 61.56 |
| 7 | Qwen 3 VL 32B (Thinking) | 59.46 |
| 8 | GLM-4.6V | 51.05 |
| 9 | Qwen 3 VL 32B Instruct | 47.45 |
| 10 | Llama 4 Scout Instruct | 43.84 |
| 11 | Llama 4 Maverick Instruct | 42.64 |
Interactive version: theaggregate.ai/benchmark?slug=mmerror-error-type-classification-mathematics-logic · How It Works · Data refreshed daily, snapshot 2026-09-25.