revns

ChatGPT Health vs revns: answers from your records or a plan you can work to

OpenAI introduced ChatGPT Health in January 2026: a health space in ChatGPT where you can link wellness apps such as Apple Health and, in the US, medical records. OpenAI says connected data is not used to train its foundation models, and ChatGPT asks before using it by default. It is built to answer health questions from your data. revns is a tracker with a goal and a plan that an assistant can work on. Its MCP server connects ChatGPT, Claude, Claude Code and Cursor to the same state: your day score, trends, goals and week plan, and, if you allow it, tools to log, set goals and plan days. It gives training guidance only, not medical advice. If you want to ask questions about lab results and records, ChatGPT Health is built for that. If you want the assistant to keep a plan with you, revns holds that state.

ChatGPT Healthrevns
Where your data livesLinked sources inside ChatGPTIn the tracker; any MCP assistant reads and writes it
Which assistantChatGPTClaude, ChatGPT, Claude Code, Cursor, any remote MCP client
Medical recordsUS hospital systems, One Medical and Function HealthNo. Training guidance only, no diagnosis
Goals and a week plan that persistNot the focusGoal pace with a range; a week plan written into each day
Daily score and morning briefNoTraining, sleep, meals and mood in one score, with a brief
PermissionsLinked sources; asks before using by defaultSeparate read and write scopes; disconnect any assistant any time

Where it is the better choice

ChatGPT Health reaches medical records and lab data that no fitness tracker does, and ChatGPT's reasoning is strong. For questions about records and results it covers ground this does not, and it is not a replacement for a clinician.

Sources

Accurate as of November 15, 2026. Training and nutrition guidance only, not medical advice. See a clinician for symptoms, injury or a medical condition.

Your first brief is one entry away.

Pick an objective, say how you eat, log one thing.