PolyReal - Raw Data Extraction: leaderboard

Metric: F1 score on the Raw Data Extraction and Analysis module of PolyReal (parsing NMR, IR spectra and raw data files); higher is better. Source: arxiv.org. Saturation forecast: Around November 2027. 15 models tracked.

Top models

#ModelScore
1Gemini 2.5 Flash0.44
2GPT-50.43
3O30.42
4Gemini 2.5 Pro0.41
5Qwen 3 VL 32B (Thinking)0.39
6Claude Sonnet 4.5 (Thinking)0.36
7Qwen 3 VL 235B A22B (Thinking)0.36
8Gemini 2.0 Flash (Thinking)0.33
9DeepSeek R10.33
10GPT-4o0.32
11Grok 40.3
12Qwen 2.5 VL 72B Instruct0.25
13GPT-4o Mini0.21

Interactive version: theaggregate.ai/benchmark?slug=polyreal-raw-data-extraction · How It Works · Data refreshed daily, snapshot 2026-10-07.