OpenAI Ships GPT-5.5 Instant to All ChatGPT Users, Claims 52% Fewer Hallucinations in High-Stakes Domains
OpenAI's newest default model rolls out over two days with major upgrades in factuality, memory, and personalization — plus hints of a new omnimodal voice mode built on the same foundation.
OpenAI began rolling out GPT-5.5 Instant on Monday as the new default model for all ChatGPT users, marking one of the company's most consequential model swaps since the original GPT-4 launch. As @OpenAI announced, the model delivers "smarter, clearer, and more personalized answers in a warmer, more natural tone" — and, critically, more concise ones. The rollout is phased over two days and extends to the API under the handle `gpt-5.5-chat-latest`, as @OpenAI confirmed in a follow-up post.
The headline number is a claimed 52.5% reduction in hallucinations across high-stakes domains. OpenAI didn't specify which domains or evaluation methodology produced that figure, but the framing suggests medical, legal, and financial queries — precisely the areas where factual reliability has been the hardest sell for enterprise adoption. If the number holds up under independent scrutiny, it would represent meaningful progress on the problem that has most stubbornly resisted scaling solutions.
Get our free daily newsletter
Get this article free — plus the lead story every day — delivered to your inbox.
Want every article and the full archive? Upgrade anytime.
No spam. Unsubscribe anytime.