GeoMMBench: leaderboard
Metric: Micro-averaged accuracy (%) on all 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 2029. 36 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 1.5 Pro | 70.7 |
| 2 | Gemini 2.0 Flash | 70.1 |
| 3 | GPT-4o | 68.5 |
| 4 | O1 | 65.8 |
| 5 | Qwen 2.5 VL 7B | 62.5 |
| 6 | Qwen 2 VL 7B | 59.8 |
| 7 | Phi-4 Multimodal Instruct | 55.3 |
| 8 | Qwen 2 VL 2B | 48.9 |
| 9 | GPT-5 | 46.1 |
Interactive version: theaggregate.ai/benchmark?slug=geommbench · How It Works · Data refreshed daily, snapshot 2026-10-07.