OpenAI has launched GPT-5.5 Instant, enhancing ChatGPT's capabilities with smarter, more personalized responses, and improving accuracy by 52.5%.
OpenAI has officially launched the GPT-5.5 Instant model, which will replace the earlier GPT-5.3 Instant in ChatGPT, making it the default neural network for users. The rollout is expected to be completed within two days.
GPT-5.5 Instant is now being introduced in ChatGPT, featuring a significant upgrade that enhances the quality of responses. Users can expect smarter, clearer, and more personalized answers delivered in a warm and natural tone. Additionally, responses are now more concise, a change that users have requested. We believe you will enjoy interacting with it. pic.twitter.com/HSQOhjqxp7
— OpenAI (@OpenAI) May 5, 2026
The GPT-5.5 Instant model shows marked improvements in accuracy, particularly in specialized fields such as medicine, law, and finance. Internal assessments reveal that it produces 52.5% fewer incorrect answers than its predecessor, with a notable 37.3% reduction in errors during complex conversations.
Furthermore, the model can autonomously decide when to access the internet for additional information, enhancing its performance on everyday tasks like image analysis and mathematical problem-solving.
Enhancements have also been made to memory and personalization features. ChatGPT now better utilizes context from previous interactions and saved data, allowing users to see how their information is being used and manage their data settings more effectively. These personalization updates are gradually being implemented for Plus and Pro plan users, with mobile app support to follow.
OpenAI's Competitive Edge in Programming
In a significant boost for its programming capabilities, OpenAI has seen a dramatic increase in downloads of its Codex solution, rising from 5 million to 86 million in just five days, surpassing downloads of competing tools such as Claude Code. In contrast, downloads of Anthropic's tool fell from 11.8 million to 7.2 million during the same period, as reported by TickerTrends.
This surge follows a major update for Codex, which now allows the AI to interact with applications on computers, including screen visibility and text input. A new built-in browser feature enables users to give precise instructions directly on web pages. Additionally, the introduction of the /goal command allows Codex to plan and execute complex tasks across multiple sessions, while the "pets" feature provides users with a way to monitor the AI agent's activity and status.
In contrast, Anthropic has faced challenges, including user reports of blocks and increased token consumption due to new features, which have affected its user base and appeal in the market.
