Language-dependent variation in observed mechanistic performance of web-enabled large language models across myeloid–mucosal immune contexts: a blinded expert evaluation
ObjectiveTo determine whether observed mechanistic performance of web-enabled large language model (LLM) responses differs across response languages under deployed web conditions and to distinguish such language-associated shifts in relative model perf…