PolyReal - Lab Safety Analysis: leaderboard

PolyReal (2026) evaluates multimodal models on 545 expert-written polymer science tasks covering five stages of real experimental work. Answers are scored against expert key points by an LLM judge and summarized as F1 from 0 to 1. This board is F1 on the module about identifying hazards in photos of real laboratories; higher is better.

Metric: F1 score on the Lab Safety Analysis module of PolyReal (risk identification in cluttered lab scenes); higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 15 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1O30.412#237
2GPT-50.373#178
3Gemini 2.5 Pro0.369#276
4Gemini 2.5 Flash0.366#463
5Qwen 3 VL 235B A22B (Thinking)0.347#459
6Claude Sonnet 4.5Thinking0.317#261
7Gemini 2.0 Flash (Thinking)0.267
8Intern-S10.249#624
9Qwen 3 VL 32B (Thinking)0.23#573
10Grok 40.224#319
11Qwen 2.5 VL 72B Instruct0.144#841
12DeepSeek R10.136#482
13GPT-4o0.117#751
14GPT-4o Mini0.078#1667
15DeepSeek-VL20.036#2927

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

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