GPT-5.5 becomes ChatGPT's default as GPT-4.5 retires
OpenAI has moved GPT-5.5 to the centre of ChatGPT, making it the default model for everyday use and retiring several older versions in the process. The company describes GPT-5.5 as its smartest model to date, with gains aimed squarely at coding, research and data analysis.
The most visible change for regular users is GPT-5.5 Instant, the new default in both ChatGPT and the API. OpenAI says it produces smarter, more accurate answers with fewer hallucinations and better personalisation controls. It has also been retuned for style: responses are meant to read more naturally, feel better paced for practical help, and lean less on the long, bullet-heavy formatting that earlier models defaulted to.
The update comes with a round of housekeeping. As of June 12, 2026, GPT-5.2 models are no longer available in ChatGPT, and any existing conversations that used them automatically continue on the matching GPT-5.5 model. GPT-4.5 is on a 30-day sunset and will be retired from ChatGPT on June 27, 2026.
OpenAI also expanded its specialist line-up. On June 3 it added new capabilities to GPT-Rosalind, a model tuned for life-sciences research with stronger biological reasoning, medicinal-chemistry knowledge, genomics analysis and experimental-workflow support.
Taken together, the changes continue a clear pattern across the major AI tools: fewer model names to choose from, smarter defaults that need less prompting, and steady pressure on anyone still relying on an older version to upgrade.