Why benchmarking clinical LLMs from OpenEvidence, Doximity is complicated
30 July 2026 · filed under 53cf4f44ec01
STAT News reports on an installment of its AI Prognosis newsletter examining how to benchmark clinical chatbots built by OpenEvidence and Doximity. The piece is described as a conversation about comparing the performance of these leading tools, alongside separate items on investor views of artificial intelligence in biopharma.
The published summary does not specify what makes such benchmarking complicated, nor does it identify who is conducting the comparison or what methodology is under discussion. The headline itself frames the task as complex, but the supporting material stops short of detailing the sources of that complexity.
STAT News attributes the reporting to its AI Prognosis newsletter, dated July 29, 2026. The outlet’s summary groups the benchmarking conversation with a separate discussion of investor sentiment toward artificial intelligence in the biopharmaceutical sector, indicating the issue covers more than one topic under a shared newsletter format.
No further detail on the benchmarking methods, results, or specific claims made by OpenEvidence or Doximity about their respective tools is available in the supplied material. The full article sits behind STAT’s subscription paywall, and the Specola’s sourcing here is limited to the outlet’s own public summary and headline.