In consultation with physicians, OpenAI said it's made improvements to GPT-5.5 Instant, the model that powers ChatGPT Health for free users.
OpenAI updates how users access ChatGPT Health OpenAI has also adjusted how individuals interface with ChatGPT Health.
At launch, OpenAI designed ChatGPT Health as a separate destination for health-related conversations, meaning users had to switch between interfaces if they were discussing health.
ChatGPT Health is still not HIPAA-compliant, or at least that is not advertised as part of OpenAI's announcement about the tool.
Users can disconnect their Health information at any time, OpenAI added, as well as disconnect Health chats from ChatGPT's memory.
OpenAI has expanded its ChatGPT Health product to all U.S. users aged 18 and older, marking the chatbot's move from its early piloting stage into the broader public.
The tool, originally announced in January of this year, allows users to connect their medical records from supported platforms to ChatGPT. This feature aims to provide more contextualized answers to users' medical questions, making conversations more informational and personalized.
OpenAI said it's spent the past few months learning from ChatGPT Health's early user base and improving the models behind the chatbot's health-related conversations. In consultation with physicians, OpenAI said it's made improvements to GPT-5.5 Instant, the model that powers ChatGPT Health for free users.
OpenAI said physician consultants have reported gains in accuracy, clarity of communication and completeness, as well as the bot's ability to follow instructions and helpfulness with health decisions.
The company also highlighted GPT-5.6 Sol, which is available to paid users. This model is OpenAI's "strongest model yet for health," the company said, stating that it is effective for addressing more complex medical queries.
OpenAI updates how users access ChatGPT Health OpenAI has also adjusted how individuals interface with ChatGPT Health. At launch, OpenAI designed ChatGPT Health as a separate destination for health-related conversations, meaning users had to switch between interfaces if they were discussing health. However, the company found that health and well-being are a part of many everyday conversations, from meal-prepping to finding a restaurant on a trip to exploring different outdoor family activities. Users may not intend to discuss health when exploring these topics, but it wasn't uncommon for ChatGPT to ask about dietary restrictions or relevant injuries during everyday chats. All said, about 70% of health-related conversations took place outside of the dedicated ChatGPT Health space. To that end, OpenAI has updated its system. Users can now give ChatGPT permission to draw on relevant information that's been connected in Health across all conversations, regardless of topic. This should prevent users from needing to take the extra step of toggling to Health when health-related questions unexpectedly arise. OpenAI said the Health tab on the left sidebar will remain a space dedicated to healthcare-specific conversations, such as getting insights into lab results or examining health trends over time. ChatGPT Health is still not HIPAA-compliant, or at least that is not advertised as part of OpenAI's announcement about the tool. The company did say that health information connected to Health is not used to train any models or target ads. OpenAI also stressed that users decide when and how ChatGPT uses information connected in Health. Users can disconnect their Health information at any time, OpenAI added, as well as disconnect Health chats from ChatGPT's memory.
