OpenVLM AI2D - Types Of: leaderboard
OpenCompass OpenVLM AI2D subcategory score for diagram questions about type classification.
Metric: Accuracy (%). Source: huggingface.co. Saturation forecast: Around January 2027. 284 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 | 88 | #98 |
| 2 | InternVL2.5-78B | 86.6 | #431 |
| 3 | Qwen 2 VL 72B | 86.3 | #460 |
| 4 | Gemini 2.5 Pro | 86 | #152 |
| 5 | GPT-4.5 | 85.7 | #274 |
| 6 | InternVL3-78B | 84.5 | #352 |
| 7 | InternVL3-38B | 84.3 | #405 |
| 8 | GPT-4o ChatGPT | 83.7 | |
| 9 | GPT-5 Mini (2025-08-07) | 82.8 | #172 |
| 10 | Gemini 1.5 Flash (002) | 82.5 | #583 |
| 11 | GPT-4.1 (2025-04-14) | 81.9 | #250 |
| 12 | Gemini 1.5 Pro (002) | 81.3 | #349 |
| 13 | InternVL3-14B | 81.3 | #505 |
| 14 | InternVL3-8B | 79.6 | #620 |
| 15 | Qwen 2.5 VL 7B | 78.1 | #643 |
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-types-of · How It Works · Data refreshed daily, snapshot 2026-10-20.