OpenAI Offers Free ChatGPT to US Clinicians, Releases HealthBench
OpenAI has made a full-featured version of ChatGPT available free to verified US medical professionals — physicians, pharmacists, and other licensed clinicians — under the ChatGPT for Clinicians brand, launched simultaneously with HealthBench Professional, an open benchmark for evaluating AI on real clinical chat tasks.
What the Source Actually Says
ChatGPT for Clinicians covers documentation and research workflows, the two highest-friction areas in clinical practice. Access is free and restricted to verified license holders in the US. The verification gate controls distribution without adding a billing barrier — a deliberate seeding strategy that bypasses the institutional procurement process and puts the tool directly into clinicians' hands.
HealthBench Professional provides the evidentiary anchor: GPT-5.4 (reported by @emollick as ChatGPT-5.4) outperforms physicians with "unlimited time and web access" on real and difficult clinical tasks. The benchmark is fully open, which enables independent replication. The credibility caveat that both @emollick and @thekaransinghal flag is significant: the benchmark was designed by OpenAI. Self-designed benchmarks on which the evaluating company's model excels warrant external validation before being cited in clinical or regulatory contexts.
A secondary finding from @emollick: GPT-5.4 Thinking and GPT-5.4 Pro produce significantly better results on complex medical image and analysis tasks compared to lower-tier models — a dependency that is "not intuitive or explained anywhere." For clinical deployments, this undocumented performance gap between model tiers matters practically: a clinician using the free tier may not be accessing the version of the model that passed the benchmark.
Strategic Take
Free distribution to licensed professionals is a well-established market-seeding strategy: remove friction, collect usage data, build domain-specific reputation, then convert institutions later. For AI strategy leaders: the HealthBench caveat is worth tracking — a credible third-party clinical benchmark evaluation would either validate the launch claim or reframe it materially.