Clair3 is a small variant caller for long-reads. Compare to PEPPER (r0.4), Clair3 (v0.1) shows a better SNP F1-score with ≤30-fold of ONT data (precisionFDA Truth Challenge V2), and a better Indel F1-score, while runs generally four times faster. Clair3 makes the best of both worlds of using pileup or full-alignment as input for deep-learning based long-read small variant calling. Clair3 is simple and modular for easy deployment and integration.