OpenAI has introduced a new feature to ChatGPT that enhances its web search capabilities through “memory.” The update, known as Memory with Search, is now available in ChatGPT and works with the latest o3 and o4-mini models. This enhancement allows the AI to learn from past user interactions, helping it deliver more personalized and relevant search results.
According to the official OpenAI Help Center, Memory with Search combines artificial intelligence’s ability to learn user preferences with real-time online search, powered by Bing or other search partners. By recalling prior conversations, ChatGPT can better understand individual interests—such as favorite foods, communication styles, or hobbies—and use this knowledge to refine its search output. For instance, if a user regularly discusses Japanese cuisine, ChatGPT may prioritize this context when suggesting restaurants, recipes, or cultural content, resulting in more relevant and useful results.
Previously, ChatGPT’s memory feature, launched in early 2024, had not been applied to online search. Its new integration into search represents a step toward more intelligent and personalized digital assistance. TechCrunch notes that Memory with Search is likely part of OpenAI’s broader strategy to distinguish ChatGPT from competitors like Anthropic’s Claude and Google’s Gemini, both of which are also developing memory-based features.
Privacy and User Control Remain Key
OpenAI highlights that users have full control over this memory feature. They can access their memory settings through the app, where they have the option to view, delete, or completely disable memory at any time. The company stresses that data is solely used to improve each individual user’s experience and is not shared with external parties.
Currently, the Memory with Search function is only available to a limited group of users, notes NIX Solutions. However, OpenAI plans to gradually roll it out to a broader audience. As with other premium features, it will be accessible to ChatGPT Plus subscribers and will start with GPT-4-turbo.
This feature marks another step forward in creating AI tools that adapt more closely to users’ needs. We’ll keep you updated as more integrations become available and access expands in the coming weeks.