OpenAI has expanded access to ChatGPT Deep Research, bringing the research-focused tool to free-tier users for the first time. The move gives non-paying ChatGPT users a limited monthly allowance while keeping the standard version available to paid tiers.
The rollout also adds a clearer split between two versions of Deep Research: the standard model-backed experience and a lighter version designed to be cheaper to serve while still preserving much of the tool’s usefulness.
What free users now get
Free ChatGPT users can now run Deep Research, but the access is limited. The free tier receives 5 tasks/month, and those tasks use the lightweight version based on the o4-mini-model.
To use it, a person has to select the Deep Research button below the chat input field before sending a prompt. That extra step matters because Deep Research is not simply the default chat mode; it is a separate research workflow that must be activated before the request is submitted.
For free users, the practical takeaway is straightforward: Deep Research is no longer only a paid feature, but it is still rationed. The monthly cap means it is best reserved for tasks that need more structured investigation than a normal chat response.
How the lightweight version differs
The standard Deep Research tool continues to run on OpenAI’s o3 model. That version had previously been restricted to paid tiers, and it remains the main Deep Research option for users with access to the higher monthly limits.
The newly available lightweight version is powered by a version of OpenAI o4-mini. According to OpenAI, it is nearly as intelligent as the existing Deep Research experience while being significantly cheaper to serve. OpenAI also says responses will typically be shorter while maintaining depth and quality.
That framing explains why the lightweight version is being used for free access and as an overflow option for some paid tiers. It allows OpenAI to offer more Deep Research availability without treating every request as a standard o3-based task.
Usage limits by ChatGPT tier
The source outlines a tiered system for Deep Research access. The limits separate standard tasks from lightweight tasks, and they vary depending on the type of ChatGPT plan.
- Free users receive 5 tasks/month using the lightweight version.
- Plus & Team users receive 10 tasks/month using the standard model, plus an additional 15 tasks/month using the lightweight version.
- Pro users receive 125 tasks/month using the standard model, plus an additional 125 tasks/month using the lightweight version.
- Enterprise users receive 10 tasks/month using the standard model.
For Plus, Pro, or Team users, the lightweight version also serves as a fallback after the monthly standard-model limit has been reached. Once those users exhaust their o3-based Deep Research tasks, additional tasks automatically use the o4-mini model until the monthly limit resets.
This means paid users do not necessarily lose Deep Research access as soon as the standard allowance is used. Instead, some tiers continue through the lightweight allocation, with shorter responses expected in exchange for continued availability.
How the reset works
Deep Research limits reset every 30 days. The cycle begins the first time a user activates Deep Research, and it then renews on that same schedule.
Users can also check how many tasks remain. The remaining-task count is available by hovering over the Deep Research button in the ChatGPT window.
That reset design makes the activation point important. The 30-day period does not simply follow a general calendar month in the description provided; it starts when the user first turns on Deep Research.
Why this matters for AI research tools
The expansion reflects a broader shift in AI products: research-style assistants are becoming part of mainstream chatbot access rather than sitting only behind premium plans. The source notes that other platforms offer similar tools, including limited free access to Deep Research in Gemini from Google, a version in Grok from xAI, and a comparable tool in Perplexity’s core product.
For users, the key change is not that every Deep Research workflow is now unlimited. It is that free ChatGPT users can try the feature directly, while paid users receive a more layered set of allowances across standard and lightweight versions.
The result is a more graduated Deep Research system inside ChatGPT. Free users get a small monthly entry point. Plus & Team users get a mix of standard and lightweight tasks. Pro users receive much higher caps across both versions. Enterprise users keep a standard-model allocation as described in the source.
In short, Deep Research is becoming more available, but access still depends heavily on tier, model version, and monthly task limits.