SciVQR - Astronomy: leaderboard

The astronomy questions of SciVQR (2026), 3,254 image-based scientific questions from competitions, entrance exams and graduate-level sources in English and Chinese, answered with zero-shot chain-of-thought. Accuracy in percent; higher is better.

Metric: Astronomy accuracy (%) with zero-shot chain-of-thought prompting on SciVQR (3,254 image-based scientific questions from competitions, entrance exams and graduate-level sources, 78 percent multiple-choice and 22 percent free-form, in English and Chinese, high-school to graduate level); higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 15 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1O174.5#403
2Gemini 2.5 Flash (Preview 04-17)72#453
3O4 Mini71#368
4GPT-4o66.5#751
5MiMo-VL-7B-RL-250866#823
6Qwen 2.5 VL 72B Instruct62.5#841
7Keye-VL-1.5-8BThinking61.5#1720
8InternVL3-78B-Instruct57.5#1415
9InternVL3-14B-Instruct55.5#1801
10InternVL3-8B-Instruct52.5#2872
11Qwen 2.5 VL 7B Instruct48#1902
12LLaVA-OneVision-7B37.5#2981
13Molmo-7B-D-092435#7207
14LLaVA-NeXT-7B29#5032
15MiniCPM-o-2.627#2423

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=scivqr-astronomy · How It Works · Data refreshed daily, snapshot 2026-11-21.