OpenVLM AI2D - Water, Carbon, and Nitrogen Cycle: leaderboard
Metric: Accuracy (%). Source: huggingface.co. 284 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5 | 77.3 |
| 2 | InternVL3-78B | 77.3 |
| 3 | InternVL3-38B | 77.3 |
| 4 | Gemini 2.5 Pro | 75 |
| 5 | InternVL2.5-78B | 75 |
| 6 | GPT-4.1 (2025-04-14) | 75 |
| 7 | InternVL3-8B | 72.7 |
| 8 | Qwen 2 VL 72B | 72.7 |
| 9 | Claude 3.5 Sonnet (20241022) | 72.7 |
| 10 | Claude 3.5 Sonnet | 72.7 |
| 11 | GPT-4.5 | 72.7 |
| 12 | GPT-4o ChatGPT | 70.5 |
| 13 | InternVL2-8B | 68.2 |
| 14 | GPT-5 Nano | 68.2 |
| 15 | Gemini 1.5 Flash (002) | 65.9 |
Interactive version: theaggregate.ai/benchmark?slug=openvlm-ai2d-water-carbon-and-nitrogen-cycle · How It Works · Data refreshed daily, snapshot 2026-09-05.