Share your thoughts, 1 month free Claude Pro on usSee more

Material science on ChemBench Material science

0.42Spearman Correlation

SCICUSTOM

Updated 14d ago

Evaluation Results

Method	Links
SCICUSTOM 2026.05		0.42	0.35
GPQA 2026.05		0.21	0.1
MedQA 2026.05		0.01	-0.09
GPT-5 2026.05		-0.04	-0.06
Embedding 2026.05		-0.39	-0.31
IfBench 2026.05		-0.43	-0.29
MMLU 2026.05		-0.61	-0.49
SimpleQA 2026.05		-0.78	-0.56