MMErroR - Error Type Classification - Earth & Environment: leaderboard

Metric: Error Type Classification Accuracy (%). Source: arxiv.org. Saturation forecast: Around December 2026. 12 models tracked.

Top models

#ModelScore
1Kimi K2.566.67
2Seed 2.0 Pro66.16
3Gemini 3 Pro (Preview)65.39
4GPT-5.2 (xHigh)65.14
5Qwen 3 VL 32B (Thinking)60.81
6Grok 459.29
7Claude Opus 4.556.74
8GLM-4.6V52.16
9Qwen 3 VL 32B Instruct36.64
10Llama 4 Maverick Instruct28.5
11Llama 4 Scout Instruct28.24

Interactive version: theaggregate.ai/benchmark?slug=mmerror-error-type-classification-earth-environment · How It Works · Data refreshed daily, snapshot 2026-09-25.