GeoMMBench - Multispectral and Hyperspectral: leaderboard

The multispectral and hyperspectral imagery questions of GeoMMBench (2026), an expert-written multimodal geoscience and remote sensing question benchmark answered zero-shot (test set). Micro-averaged accuracy in percent; higher is better.

Metric: Micro-averaged accuracy (%) on the multispectral and hyperspectral imagery questions of the GeoMMBench test set (expert-written multimodal geoscience and remote sensing multiple-choice and numeric questions, zero-shot, LMMs-Eval, rule-based answer extraction, invalid outputs counted wrong); higher is better. Source: arxiv.org. Saturation forecast: Around June 2027. 35 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Qwen 3 VL 8B75.7#917
2Qwen 3 VL 4B73#1151
3Qwen 2.5 VL 7B70.3#1862
4Phi-4 Multimodal Instruct70.3#4154
5InternLM-XComposer2.570.3#3817
6Qwen 3 VL 30B70.3#561
7Qwen 2 VL 7B67.6#2275
8NVILA-8B67.6#1977
9GPT-4o64.9#751
10Gemini 2.0 Flash64.9#723
11Gemini 1.5 Pro64.9#1063
12Qwen 2.5 VL 3B62.2#2639
13CogVLM262.2
14O159.5#403
15LLaVA-OneVision-7B59.5#2981

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=geommbench-multispectral-and-hyperspectral · How It Works · Data refreshed daily, snapshot 2026-11-21.