OpenVLM AI2D - Eclipses: leaderboard
OpenCompass OpenVLM AI2D subcategory score for diagram questions about eclipses.
Metric: Accuracy (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 284 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 100 | #152 |
| 2 | InternVL3-78B | 100 | #352 |
| 3 | Qwen 2 VL 72B | 100 | #460 |
| 4 | InternVL3-38B | 96.8 | #405 |
| 5 | InternVL2.5-78B | 96.8 | #431 |
| 6 | Moonshot v1 8K | 96.8 | #617 |
| 7 | GPT-5 | 93.5 | #98 |
| 8 | InternVL3-8B | 93.5 | #620 |
| 9 | Gemini 1.5 Flash (002) | 93.5 | #583 |
| 10 | GPT-4.1 (2025-04-14) | 93.5 | #250 |
| 11 | Aquila-VL-2B | 93.5 | #1238 |
| 12 | Qwen 2.5 VL 7B | 93.5 | #643 |
| 13 | Grok 2 (1212) | 93.5 | #496 |
| 14 | GPT-4o ChatGPT | 93.5 | |
| 15 | Claude 3.5 Sonnet (20241022) | 90.3 | #291 |
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-eclipses · How It Works · Data refreshed daily, snapshot 2026-10-20.