OpenAI launches ChatGPT Health with medical records integration
Jan 7, 2026
Key Points
- OpenAI launched ChatGPT Health, a private space where users connect medical records and wellness apps like Apple Health to receive personalized health guidance and appointment preparation.
- OpenAI claims 40 million people globally ask ChatGPT health questions daily, with roughly 20% of all queries health-related, positioning the platform as a durable alternative to fragmented lab-result portals.
- ChatGPT Health faces pressure from telehealth startups like Doctoronic that employ physicians, forcing OpenAI to choose between building a clinical network or remaining vulnerable to foundation model improvements that could obsolete the product layer.
Summary
OpenAI launched ChatGPT Health, a dedicated private space where users can connect medical records and wellness apps including Apple Health and Peloton. Users input blood work and health data to get personalized support for doctor appointment prep, meal planning, and exercise routines tailored to their conditions. Access is by request rather than immediate rollout.
OpenAI reports over 40 million people globally ask ChatGPT health questions daily, with roughly 20% of all ChatGPT queries health-related. The product positions ChatGPT as a consumer health destination at scale. Unlike fragmented lab-result portals from Quest or LabCorp that degrade over time, importing health records into ChatGPT creates stickiness because OpenAI is unlikely to disappear.
The product faces two competitive pressures. Doctoronic launched recently and is building toward full telehealth by employing doctors, writing prescriptions in Utah, and running referrals. OpenAI is unlikely to employ physicians itself and would more likely build a network around its platform. Speed matters here: how fast ChatGPT Health moves into clinical territory will determine whether Doctoronic's head start matters.
There is also a classic AI-wrapper risk. If ChatGPT Health depends on models improving at long context, multi-page answers, or math, the underlying model improvements will obsolete the product layer. The safer play is building something unique, whether domain-specific privacy, network effects, or structural defensibility that doesn't rely on the foundation model staying frozen.
One speculative angle: a legal version with attorney-client privilege. People already use ChatGPT for legal work on standard subscriptions. A premium tier with privilege protections could command $100 per month, though licensing and liability questions remain unsolved.