Overview
- OpenAI announced Thursday that GPT-5.5 Instant is now the default model for all ChatGPT users, replacing the prior Instant variant and extending the company’s highest-rated health-answering capability to the free tier.
- The company reports large improvements on its in-house HealthBench tests and production monitors, including a 52.5% drop in hallucinated claims on high‑stakes prompts and a 71% fall in health responses flagged for factual issues over two months.
- OpenAI says a global panel of more than 260 physicians reviewed roughly 700,000 example responses and rated GPT-5.5 Instant higher than physician-written answers across accuracy, communication, and completeness in a 3,500-sample comparison.
- Those performance claims rest on OpenAI’s internal benchmark and physician evaluations rather than independent, peer‑reviewed testing, leaving outside verification of the reported gains unavailable.
- Practical limits and next steps matter: free users get a 16K token context window while premium subscribers can access up to 128K tokens, and wider free access could reduce clicks to publishers, increase reliance on ChatGPT for medical advice, and raise urgent verification and safety questions for clinicians and patients.