TL; DR
- OpenAI has made GPT-5.5 Instant the new default ChatGPT model for all users.
- Hallucinations dropped by 52.5% in high-stakes medical, legal, and financial queries.
- Responses are now shorter, less emoji-heavy, and more contextually personalized.
OpenAI didn’t make a big deal out of it, but the company pushed out a new default model for ChatGPT users across the globe: GPT-5.5 Instant instead of the earlier GPT-5.3 (in place since March).
The update is free, it’s already live, and it fixes some of the persistent complaints ChatGPT users have had about the chatbot for a very long time.
Also Read: CMF Watch 3 Pro Hits India: Dual-Band GPS, ChatGPT, and a 13-Day Battery Under ₹8,000
GPT-5.5 Availability
Paid subscribers on the ChatGPT Plus and Pro plans get first access to the upgraded personalization features on the web.
Other tiers, including Free, Go, Business, Enterprise, and mobile users, will get access in the coming weeks. Paid users who wish to stick with the legacy model have three months to set up their database on the new one before OpenAI completely retires it.
Also Read: Best Smartphones To Buy Under Rs. 25,000 During Flipkart SASA LELE Sale 2026
What’s New With GPT-5.5?
The headline improvement here is accuracy. OpenAI’s internal testing suggests that GPT-5.5 Instant generated 52.5% lesser hallucinated claims, particularly when compared to GPT 5.3 Instant on high-stakes prompts, such as those related to medicine, law, and finance.
Further, the newer model also cuts inaccurate claims by 37.3% in conversations that users had previously flagged for factual errors. In my opinion, those are meaningful numbers across a variety of important domains.
On the rigorous mathematics benchmark AIME, the model now scores just above 80%, a considerable jump from its predecessor’s mid-60% score. On the benchmark that tests PhD-level scientific reasoning, it reaches the mid-80% range as well.
Also Read: LG India Bets Big on Affordable-Premium and High-End Appliances With Its 2026 Home Range Launch
Then there’s the tone that users had been complaining about on forums and social media. GPT-5.5 has dialed back on its response length (responses are now about 30% shorter) and contain 29% fewer lines. Users should also come across less repetition, fewer unsolicited follow-up questions, and see fewer emojis in responses.
Regarding personalization, the model now draws more actively from your past conversations, uploaded files, and connected Gmail account to fine-tune its tone and language. Further, a new memory source feature lets you see exactly which context the chatbot used to generate a response.
You can follow Smartprix on Twitter, Facebook, Instagram, and Google News. Visit smartprix.com for the latest tech and auto news, reviews, and guides.
