Longevity Bench is a benchmark suite for evaluating large language models on quantitative aging-biology tasks. Each prompt asks a model to make a prediction from a real biomedical profile (DNA methylation, gene expression, plasma proteomics, or clinical measurements) and produce a structured answer that can be scored with a deterministic metric.
The released evaluation split covers 25 tasks across five biological domains — epigenomics, transcriptomics, proteomics… See the full description on the dataset page:
https://huggingface.co/datasets/insilicomedicine/longebench.
