Sci-Rho - Chemistry (English): leaderboard

Metric: Average-case accuracy (%) on the English chemistry templates of Sci-Rho: 606 expert-written executable templates per language (mathematics, physics, chemistry, biology and computer science, many drawn from olympiad problems), each rendered as 10 visually and numerically varied image-grounded instances; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 17 models tracked.

Top models

#ModelScore
1GPT-5.4 (Non-reasoning)94.4
2Gemini 2.5 Flash (Non-reasoning)92.2
3Gemini 2.5 Pro91.7
4Qwen 3.5 27B (Non-reasoning)90.9
5Qwen 3.5 122B A10B (Non-reasoning)88
6Llama 4 Scout Instruct84.7
7Qwen 3 VL 8B Instruct83.6
8InternVL3.5-8B79.3
9Gemma 3 12B (IT)60.1
10Molmo2-8B56.6
11Gemma 3 4B (IT)43.4
12Qwen 3 VL 4B Instruct35.8

Interactive version: theaggregate.ai/benchmark?slug=sci-rho-chemistry-english · How It Works · Data refreshed daily, snapshot 2026-09-29.