AI 导读
Sante 在医学知识和推理方面尤为突出。 每 token 仅 5.1B 激活参数，它在 MedXpertQA-Text、DiagnosisArena-MCQ 和 AFUMED-Drug 上位列 flash 级模型第一，得分分别为 53.88、83.83 和 89.59。 它在 HealthBench Professional 上达到 45.73，并在众多关键医疗基准上展现出与旗舰模型相当的性能。
正文
Sante is particularly strong in medical knowledge and reasoning.
With only 5.1B active parameters per token, it ranks #1 among flash-size models on MedXpertQA-Text, DiagnosisArena-MCQ, and AFUMED-Drug, scoring 53.88, 83.83, and 89.59, respectively.
It also reaches 45.73 on HealthBench Professional and delivers performance comparable to flagship models across many key healthcare benchmarks.
来源：@AntLingAGI · x.com