OpenVLM AI2D - Volcano: leaderboard

OpenCompass OpenVLM AI2D subcategory score for diagram questions about volcanoes.

Metric: Accuracy (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 284 models tracked.

Top models

#ModelScoreOverall rank
1Gemini 2.5 Pro100#152
2InternVL3-8B100#620
3Gemini 1.5 Pro (002)100#349
4InternVL3-38B100#405
5GPT-4.1 (2025-04-14)100#250
6Qwen 2 VL 7B100#750
7Claude 3.7 Sonnet100#249
8Gemma 3 27B100#607
9Qwen 2 VL 72B100#460
10Gemini 1.5 Pro100#445
11GPT-4.5100#274
12GPT-4o ChatGPT100
13GPT-593.8#98
14GPT-4.1 Mini93.8#355
15Gemini 2.0 Flash93.8#342

No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.

Interactive version: theaggregate.ai/benchmark?slug=openvlm-ai2d-volcano · How It Works · Data refreshed daily, snapshot 2026-10-20.