OpenAI Rolls Out GPT-5.5 Instant as ChatGPT's New Default: 52.5% Fewer Hallucinations
Original: OpenAI Rolls Out GPT-5.5 Instant as ChatGPT's New Default: 52.5% Fewer Hallucinations View original →
GPT-5.5 Instant: ChatGPT's New Default
On May 5, 2026, OpenAI launched GPT-5.5 Instant as ChatGPT's new default model, replacing GPT-5.3 Instant. The upgrade delivers smarter, clearer, and more personalized responses in a warmer, more natural tone across all ChatGPT users.
Key Improvements
- 52.5% fewer hallucinations: On high-stakes prompts covering medicine, law, and finance, GPT-5.5 Instant produced 52.5% fewer hallucinated claims than its predecessor. Inaccurate claims on conversations previously flagged by users for factual errors dropped by 37.3%.
- Conciseness: Responses are tighter and more to-the-point, with gratuitous emojis and excessive follow-up questions reduced without losing the model's personality.
- Enhanced multimodal capability: Improvements in analyzing images, answering STEM questions, and determining when to use web search for better answers.
- Personalization: Can use context from past chats, uploaded files, and connected Gmail. Rolling out to Plus and Pro users on web first, with mobile and other tiers to follow.
Benchmark Performance
GPT-5.5 Instant scored 81.2 on AIME 2025 math tests (up from 65.4) and 76 on the MMMU-Pro multimodal benchmark (up from 69.2), representing measurable gains across both reasoning and vision tasks.
Availability
GPT-5.5 Instant is the new default for all ChatGPT users and accessible in the API as chat-latest. Paid users can continue to access GPT-5.3 Instant through model settings for three months before it is retired.
Related Articles
GPT-5.5 Instant is now ChatGPT's default model, replacing GPT-5.3 Instant. The update cuts hallucinated claims by 52.5% on high-stakes medical, legal, and financial prompts, and adds Gmail-based personalization and memory-source transparency.
OpenAI announced GPT-5 on 2025-08-07 for both ChatGPT and API usage. The launch highlights include a reported 45% hallucination reduction vs GPT-4o and major benchmark gains such as HealthBench Hard 44.6.
OpenAI said on January 29, 2026 that ChatGPT would stop offering GPT-4o and older model options from February 13, 2026. GPT-4o, GPT-4.5, and o4-mini are being replaced by GPT-5, GPT-5 thinking, and o5-mini respectively.
Comments (0)
No comments yet. Be the first to comment!