OpenAI has rolled out a quiet but significant upgrade to ChatGPT’s web search capabilities, improving response quality, instruction-following accuracy, and support for more complex and longer queries. Announced via its June 13 changelog, the update brings notable refinements to how ChatGPT handles user intent and context, even in multi-turn conversations.
A key advancement includes the ability to analyze images: users can now upload a photo, and ChatGPT will extract relevant information from public websites based on the image and any accompanying query.
Additionally, the AI can now conduct multiple searches for a single question to provide more comprehensive answers, reducing redundancy and enhancing relevance in longer chats. The upgraded memory function, integrated with the ChatGPT “Projects” feature, also lets the model reference past interactions for more tailored responses.
Despite these improvements, OpenAI acknowledged some limitations. Users might encounter longer response times and unnecessarily detailed “chain-of-thought” reasoning for simple queries — an issue the company plans to address.
The update builds on the native web search engine launched in October 2024, continuing OpenAI’s effort to expand ChatGPT’s functionality with tools like downloadable Canvases and deeper project integration.