Cao, G.; Shen, Y.; Wu, J.; Chao, H.; Chen, M.; Chen, D.

We present CellReasoner, a lightweight, open-source large language model (LLM) tailored for single-cell type annotation. We introduced a compact training strategy that activates the reasoning capabilities of 7B-parameter LLMs using only 380 high-quality chain-of-thought exemplars. CellReasoner directly maps cell-level gene expression profiles to cell type labels, exhibiting robust zero- and few-shot generalization. The model further demonstrates expert-level, marker-by-marker reasoning, enabling structured, interpretable annotations and offering a practical solution for intelligent single-cell analysis.