Two months ago, OpenAI introduced its Deep Research AI agent, designed to tackle complex, multi-step research tasks by sifting through huge amounts of information online. Then, weeks later, it followed up by expanding access to this tool beyond its top-tier Pro users, making it available to everyone paying for a ChatGPT subscription, including Plus, Team, Edu, and Enterprise plans, with varying monthly usage limits depending on the tier.
Now, OpenAI is rolling out a new lightweight version of this tool. Deep Research, for those who do not know, is essentially an advanced AI agent that can conduct in-depth investigations online, acting somewhat like a digital research analyst. It finds, analyzes, and synthesizes information from potentially hundreds of web sources, aiming to produce comprehensive reports with citations.
Deep Research uses a version of its o3 model specifically tuned for browsing and data work, designed to leverage strong reasoning capabilities to handle nuanced queries and find niche details, sometimes taking significant time to complete a task.
This new lightweight version is powered by a different engine, a version of the o4-mini model. OpenAI says this makes it cheaper to run, which in turn allows them to increase the overall number of research tasks users can perform. While the company claims the lightweight version maintains much of the "depth and quality" of the original, responses will "typically be shorter."
OpenAI is adding these lightweight tasks to the existing limits for the original Deep Research tool. The AI lab confirmed this rollout initially for ChatGPT Plus, Team, and Pro users, and also made the lightweight version available for free users immediately. Access for Enterprise and educational users is expected to follow shortly after, with the same usage levels as Team users.