ChatGPT Launches New Shopping Research Agent, Personalizing the Buying Guide

OpenAI has announced the launch of a specialized Shopping Research feature within ChatGPT, engineered to fundamentally simplify the process of product selection and gift buying. This is far more than a simple search update; it is a sophisticated agent designed to take over the most time-consuming aspects of modern commerce—deep analysis and comparison. The core technological advancement lies in the use of a specialized version of the GPT-5 mini model, which has undergone extensive reinforcement learning tailored specifically for accurate commerce tasks. In internal evaluations, OpenAI found that this focused training allows the model to outperform other advanced models in the ChatGPT family, such as GPT-5-Thinking, when it comes to product accuracy in detail-heavy shopping queries. The system is rigorously trained to find, read, and synthesize information by prioritizing known, trusted retail and review sources across the web.

For the user, the shopping experience transforms into a highly personalized and interactive dialogue. Instead of sifting through dozens of review pages, the user can pose complex, natural language demands—such as, “Find the quietest vertical vacuum cleaner under $300 that still includes a HEPA filter.” The model then acts as an active consultant, leveraging its ChatGPT Memory and past conversations to personalize the search by asking crucial clarifying questions about budget, preferred features, or aesthetic needs. The process features a dynamic feedback loop, where the AI presents suggested products and allows users to guide the research along the way by providing real-time feedback, thus rapidly refining the results.

The output provided by the agent is not a mere list of links; users receive a personalized buyer's guide complete with comprehensive comparisons, detailed trade-offs between top products, and up-to-date pricing information. OpenAI has emphasized that this feature is built on transparency and trust: the research results are entirely organic and based on publicly available data, reading product pages directly and prioritizing verified sources while avoiding low-quality or spammy sites. Furthermore, the company maintains a strict policy of not sharing user chat data with retailers. While the tool guides the decision, users complete their purchase by clicking through to the retailer’s site. The feature is currently rolling out to all logged-in ChatGPT users—including those on the free plan—across mobile and web, making sophisticated, personalized research accessible to millions in time for the holiday shopping season.