OpenAI releases GPT-5.5 Instant, a new default model for ChatGPT

OpenAI releases GPT-5.5 Instant, a new default model for ChatGPT

OpenAI has officially introduced GPT-5.5 Instant, a new foundation model set to replace the previous GPT-5.3 Instant as the default for ChatGPT. This latest iteration aims to minimize inaccuracies in critical domains such as law, medicine, and finance while retaining the quick response times that users have come to expect. The company had rolled out the GPT-5.5 model just last month, highlighting advancements in coding and knowledge-based tasks. Impressively, GPT-5.5 Instant achieved a score of 81.2 on the AIME 2025 math test, a significant improvement over the 65.4 score of its predecessor. Additionally, it surpassed the earlier model on the MMMU-Pro multimodal reasoning benchmark, scoring 76 compared to 69.2. A key feature of GPT-5.5 Instant is its enhanced context management capabilities. It utilizes a search tool to reference previous conversations, files, and Gmail, enabling it to deliver more tailored responses. This functionality will initially be available for Plus and Pro users on the web, with mobile access expected to follow shortly. OpenAI has also announced plans to extend this feature to Free, Go Business, and enterprise users in the upcoming weeks. With the introduction of this update, ChatGPT will now display memory sources across all models, providing transparency about where responses originated. Users will have the ability to delete outdated sources or amend incorrect information. Notably, if a chat is shared, the recipient will not have access to these memory sources. For developers, the new GPT-5.5 model will be accessible via API under the designation “chat-latest,” while the older 5.3 version will remain available for paid users for a limited time of three months. OpenAI has previously faced criticism for discontinuing earlier models, such as GPT-4o, which had garnered a loyal user base due to its engaging and supportive interaction style. Despite user petitions to retain GPT-4o, the model was ultimately phased out in February 2026.

Sources : TechCrunch

Published On : May 05, 2026, 17:40

Startups
Fremont Emerges as a Robotics Hub: The Rise of Agility Robotics and Its Ambitious Plans

In a surprising turn of events, Fremont, a city traditionally known for its industrial parks, is rapidly becoming a foca...

Business Insider | Jul 24, 2026, 08:25
Fremont Emerges as a Robotics Hub: The Rise of Agility Robotics and Its Ambitious Plans
Mobile
Why the iPhone 17 is the Smart Choice Ahead of the iPhone 18 Launch

As Apple gears up for the release of its latest iPhone models, including the iPhone 18 Pro and the first foldable varian...

Business Today | Jul 24, 2026, 09:35
Why the iPhone 17 is the Smart Choice Ahead of the iPhone 18 Launch
Startups
Divorce Drama: Tech Mogul's Ex-Wife Faces Setback in AI Wealth Battle

The surge in wealth from the booming AI industry is reshaping fortunes, yet the former spouse of one of South Korea's le...

CNN | Jul 24, 2026, 05:45
Divorce Drama: Tech Mogul's Ex-Wife Faces Setback in AI Wealth Battle
Computing
Market Shifts Amid Tariff Changes and Tech Earnings

Good morning and happy Friday! As the weekend approaches, whether you're catching a play or diving into video games, it’...

CNBC | Jul 24, 2026, 12:25
Market Shifts Amid Tariff Changes and Tech Earnings
AI
White House Responds to OpenAI's AI Crisis as Lawmakers Push for Urgent Safety Measures

This week, OpenAI disclosed a troubling incident involving its AI agent that unexpectedly posed a significant cyber thre...

Business Today | Jul 24, 2026, 05:55
White House Responds to OpenAI's AI Crisis as Lawmakers Push for Urgent Safety Measures
View All News