Clinic Shortlists and Sources Across ChatGPT Models and Reasoning Efforts

Clinic Shortlists and Sources Across ChatGPT Models and Reasoning Efforts is a descriptive stability benchmark built from 450 core answers: three exact NYC commercial prompts across five ChatGPT model/reasoning configurations, with 30 isolated executions per cell. Fifteen technical preflight answers are excluded from all reported denominators.

Headline within-configuration provider-set overlap was 80.3% for IVF, 34.7% for full-arch dental implants, and 53.9% for bariatric programs; corresponding visible-domain overlap was 51.1%, 28.3%, and 44.8%. The visible-source core contains 3,757 markdown-link occurrences, 3,690 unique answer-by-URL pairs, and 3,194 answer-by-domain pairs.

Collection used Codex CLI 0.147.0 authenticated through one ChatGPT login with live search. It did not use the consumer ChatGPT Free/Plus UI or API-key billing. Position means presentation order, not quality ranking. Reader-visible domains do not establish causal recommendation sources. Physical US IP/location was not controlled.

Boris Teplyakov reviewed the methodology as SEO Lead. This was not clinical review, peer review, a formal external audit, or an assessment of named providers.

Article and full methodological context: https://rotgar.com/medical/resources/chatgpt-clinic-shortlists-model-effort-stability