A blinded assessment published in Discover Informatics had eight multidisciplinary reviewers rate 400 synthetic clinical outputs generated by five locally deployed open-weight models, finding that while the panel judged the primary action acceptable in a majority of cases, one in seven outputs drew a majority verdict of unsafe or insufficiently urgent, and nearly every output that fell below a clinician-defined reference range was flagged as concerning. The study generated a fresh set of 44 scenarios covering self-triage, counterfactual triage in base and worsened forms, synthetic reference-passage tasks, and contraindication or sequencing checks, with no output records reused from an earlier six-case study published in Cureus. The design called for 1584 candidate attempts from crossing 44 scenarios with six source configurations, three prompt conditions and two repeated generations; one source configuration failed technically and was excluded before sampling, leaving 1320 eligible outputs across all 44 scenarios and five retained sources, from which 400 distinct outputs were selected at random in Python. The five retained sources were later unmasked as DeepSeek-R1 14B, Gemma 3 12B, gpt-oss 20B, Llama 3 and Gemma 4 26B, all run locally through Ollama, with model identity withheld from raters. Before expert review, an independent physician-researcher with more than 20 years of clinical experience defined lower and upper reference actions for each scenario, never shown to reviewers. Each of the eight reviewers rated all 400