OpenAI Rolls Out GPT-5.5 Instant as Default ChatGPT Model
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- OpenAI launched GPT-5.5 Instant as ChatGPT's new default model, replacing GPT-5.3 Instant for all users, including free-tier users who have no other model options
- GPT-5.5 Instant is pitched as smarter, more accurate, and more concise, with OpenAI explicitly citing that ChatGPT is now 'half as likely to mislead you on medical and financial questions' (per Digital Trends' headline)
- Benchmark gains cited by testers show PhD-level science accuracy jumping from 78.5% to 85.6% and competition math rising from 65.4% to 81.2% versus the prior Instant model
- Sam Altman called the combined gains in speed, intelligence, personality, and memory 'a more-than-sum-of-the-parts thing' and urged users stuck on thinking-mode models to try Instant
- OpenAI also upgraded memory and personalization, letting ChatGPT pull context from saved memories, past chats, files, and connected Gmail accounts, with new UI showing which memory sources were used
- Michelle Pokrass, an OpenAI engineering lead, said the team focused on factuality, 'crushing hacks,' and raising baseline intelligence, calling 5.5 'a pretty big step forward on all three'
Why it matters: GPT-5.5 Instant becomes the model every ChatGPT user — including the hundreds of millions on the free tier who have no other option — hits by default, so the 7-point jump in PhD-science accuracy and the roughly 16-point jump in competition math translate into real-world changes for the largest user base in consumer AI. The tighter factuality framing also signals OpenAI is competing on trust in high-stakes domains (medicine, law, finance) rather than just raw capability.
