OpenAI Introduces GPT-5.5 Instant

OpenAI is massively upgrading its flagship product, ChatGPT, with the simultaneous rollout of the new model and personalized features, marking an important step for the company in enhancing user experience and competitiveness.

GPT-5.5 Instant will be launched as the default model for all ChatGPT users within the next two days, and will also be made available to API developers under the interface name "gpt-5.5-chat-latest".

OpenAI stated that this is a significant upgrade, with the new model capable of providing smarter, clearer, and more naturally warm responses, while also being more concise in its responses.

At the same time, OpenAI announces comprehensive improvements to memory and personalization features. The Memory Sources feature will be made available to all web users on consumer subscription packages, with mobile support launching soon; personalization enhancements will be rolled out first to Plus and Pro users, with mobile support to follow shortly.

The combination of these features aims to enable ChatGPT to better understand and serve individual users in conversations, likely increasing user stickiness.

GPT-5.5 Instant: More Reliable, Smarter, More Natural

Instant has become more reliable, significantly improving factual accuracy across various aspects, particularly in areas where accuracy is paramount, where the improvements are the most noticeable.

In internal evaluations, GPT-5.5 Instant saw a 52.5% reduction in false statements compared to GPT-5.3 Instant in high-risk prompts covering areas such as medicine, law, and finance. Additionally, it reduced inaccurate statements by 37.3% in difficult conversations marked by users as containing factual errors.

GPT-5.5 Instant is a smarter model that performs better in everyday tasks, including improved analysis of photos and image uploads, answering STEM (Science, Technology, Engineering, Mathematics) related questions, and deciding when to use internet searches to provide more useful answers.

These advancements are reflected in the increased scores across visual reasoning, mathematics, and science assessments:

After this update, ChatGPT's response is more concise and clear, with highlighted points while maintaining the substantive content, and retains its friendly and approachable personality, making it more enjoyable to use.

It can provide the same information, but it is generally more practically valuable than the previous models, while reducing lengthy responses and overly formatted issues. Additionally, it reduces unnecessary follow-up questions and avoids elements such as the misuse of emojis that could make responses seem cluttered.

Comprehensive Upgrade of Memory and Personalization Features

This update also significantly expands ChatGPT's personalization capabilities. ChatGPT is now able to more effectively invoke saved memories, historical conversation records, uploaded files, and connected Gmail accounts to provide more contextually relevant answers.

The newly launched Memory Sources feature will clearly present which contextual information is invoked in each response, allowing users to update, delete, or disconnect relevant data connections at any time, maintaining proactive control over the use of personal information.

The personalization enhancement feature