OpenVLM ScienceQA Test: leaderboard
OpenCompass OpenVLM test split results for ScienceQA: multimodal science questions spanning natural science, language science, and social science with diagram-grounded reasoning.
Metric: Accuracy (%). Source: scienceqa.github.io. Source: huggingface.co. Saturation forecast: Estimated already saturated. 218 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | InternVL3-38B | 99.3 | #405 |
| 2 | InternVL3-78B | 98.7 | #352 |
| 3 | InternVL3-14B | 98.6 | #505 |
| 4 | InternVL3-8B | 97.9 | #620 |
| 5 | InternVL2-8B | 97.1 | #817 |
| 6 | MiniCPM-V-2.6 | 96.7 | #836 |
| 7 | Aquila-VL-2B | 95.1 | #1238 |
| 8 | GPT-4.1 (2025-04-14) | 92.8 | #250 |
| 9 | Qwen 2 VL 72B | 91.2 | #460 |
| 10 | Claude 3.7 Sonnet | 90.9 | #249 |
| 11 | Grok 2 (1212) | 90.4 | #496 |
| 12 | Qwen 2.5 VL 7B | 89 | #643 |
| 13 | Claude 3.5 Sonnet | 88.9 | #350 |
| 14 | Gemma 3 27B | 87.7 | #607 |
| 15 | Pixtral-12B | 87.2 | #808 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=openvlm-scienceqa-test · How It Works · Data refreshed daily, snapshot 2026-10-20.