It’s April 2025, and OpenAI is shaking things up again. The folks behind ChatGPT just dropped a bombshell: a new “lightweight” version of their Deep Research tool is now available—not just for their premium subscribers, but for free users too. This isn’t just a minor update; it’s a bold move to make advanced AI research tools more accessible while keeping costs in check.

If you’re new to the ChatGPT universe, Deep Research is like having a tireless research assistant who scours the internet, sifts through mountains of data, and churns out detailed, cited reports in minutes. Launched earlier this year, it’s powered by a specialized version of OpenAI’s o3 model, designed for heavy-duty tasks like competitive market analysis, scientific literature reviews, or even helping you decide which car to buy by comparing specs and reviews across dozens of sources. Unlike ChatGPT’s quick-fire responses, Deep Research takes its time—anywhere from 5 to 30 minutes—delivering structured, fully documented reports that rival the work of a human analyst.

But here’s the catch: the original Deep Research was a resource hog. Running on the beefy o3 model, it was expensive to operate, which meant OpenAI initially limited it to Pro users (those shelling out $200 a month) and later expanded it to Plus and Team subscribers with strict usage caps. Free users? They were left out in the cold. That is, until now.

On April 24, 2025, OpenAI announced the rollout of a “lightweight” Deep Research tool, powered by their recently launched o4-mini model. This isn’t just a stripped-down version; OpenAI claims it’s “nearly as intelligent” as the full-fledged Deep Research while being significantly cheaper to run. The trade-off? Responses are shorter, but OpenAI insists they still pack the same depth and quality, complete with citations and structured logic. Think of it as the difference between a 20-page research paper and a 5-page executive summary—both get the job done, but one’s easier to digest.

The lightweight version is a strategic move for OpenAI. By leaning on the more efficient o4-mini model, the company can scale access without breaking the bank. As OpenAI noted in a post on X, this allows them to “up usage limits” across the board. Free users now get five Deep Research tasks per month using the lightweight version. Plus and Team subscribers score 10 standard tasks plus 15 lightweight ones, while Pro users get a whopping 125 standard and 125 lightweight tasks. Enterprise and Education users will join the party next week, matching the Team tier’s limits. If you hit your cap on the standard version, ChatGPT seamlessly switches to the lightweight model, ensuring you’re never left hanging.

Why does this matter? OpenAI reportedly spends “tens of millions of dollars” processing basic user interactions (yes, even your polite “please” and “thank you”s add up). A cheaper-to-run model means they can democratize access to powerful tools without burning through cash. It’s a win-win: users get more research firepower, and OpenAI keeps its servers humming.

Using the lightweight Deep Research tool is straightforward. In ChatGPT’s web interface (mobile and desktop apps are coming soon), you select “Deep Research” from the message composer, type your query, and optionally upload files like PDFs or spreadsheets for extra context. Want a report on the best electric vehicles for 2025? Need a competitive analysis of cloud computing providers? Just ask, and the tool will comb the web, analyze sources, and deliver a concise, cited report. For example, a query like “Compare the top translation apps for 2025” might yield a report summarizing features, pricing, and user reviews, all backed by links to reputable sources.

The lightweight version maintains the core strengths of its predecessor: multi-step reasoning, real-time web browsing, and the ability to parse text, images, and documents. OpenAI’s planning to sweeten the deal with future updates, like adding charts, data visualizations, and access to premium or private data sources. For now, though, it’s all about making high-quality research accessible to more people.

OpenAI isn’t alone in the AI research game. The past year has seen a flurry of “deep research” tools from tech giants and startups alike. Google’s Gemini rolled out its own Deep Research feature, leveraging Google Search for real-time data retrieval and faster reports (5–10 minutes compared to OpenAI’s 5–30). xAI’s Grok 3 offers a “DeepSearch” mode for subscribers, boasting similar web-scraping chops. Microsoft’s Copilot, Perplexity AI, and Anthropic’s Claude have also jumped into the fray, each with their own take on agentic research tools.

What sets OpenAI’s lightweight Deep Research apart is its accessibility. While Google made its Deep Research free for all Gemini users in March, OpenAI’s move to include free ChatGPT users (albeit with a five-task limit) is a big deal. It’s a clear signal that OpenAI wants to stay competitive in a crowded market, especially as rivals like China’s DeepSeek gain ground with cost-effective models like R1.