Overview
- OpenAI has rolled out GPT-5.5 Instant as the default ChatGPT model and says the update is available to all free and paid users.
- The company asserts GPT-5.5 Instant now matches its most advanced 'Thinking' models on difficult health evaluations, calling this parity on medical tasks.
- StartupHub.ai reported a 71% reduction in factuality issues after the update while Crypto Briefing cited OpenAI figures showing a 52.5% drop in hallucinated claims and a 37.3% fall in user-flagged errors, and these numbers come from OpenAI’s internal testing.
- OpenAI says the gains were driven by underlying model advances plus a physician-led evaluation process that included more than 260 clinicians across roughly 60 countries who helped define and score good health responses.
- The company points to broad reach—StartupHub.ai reports about 230 million weekly health interactions—to argue the changes matter for public health, but the coverage notes that the accuracy and testing claims are company-reported and have not been independently verified.