OpenVLM AI2D - Rock Cycle — leaderboard

Metric: Accuracy (%). Source: huggingface.co. 284 models tracked.

Top models

#ModelScore
1GPT-5 (2025-08-07)82.1
2Qwen 2 VL 72B79.1
3InternVL2.5-78B79.1
4GPT-4.1 (2025-04-14)79.1
5InternVL3-78B74.6
6InternVL3-8B74.6
7GPT-4o ChatGPT74.6
8GPT-4.574.6
9GPT-5 Mini (2025-08-07)74.6
10Gemini 2.5 Pro73.1
11InternVL2-8B73.1
12InternVL3-38B73.1
13Qwen 2 VL 7B71.6
14MiniCPM-V-2.671.6
15Gemini 1.5 Pro (002)70.1

Interactive version: theaggregate.ai/benchmark?slug=openvlm-ai2d-rock-cycle · How the rankings work · Data refreshed daily, snapshot 2026-07-22.