If you’ve ever wanted ChatGPT to go beyond quick answers and conduct in-depth research, there’s good news—OpenAI is now rolling out the Deep Research feature to all paid users. This includes ChatGPT Plus, Team, Enterprise, and Education subscribers. Previously, this feature was exclusive to the expensive Pro tier at $200 per month.
But what exactly is this Deep Research, and how can you use it? More importantly, how does it compare to similar tools from Google Gemini, Perplexity, and Grok? Let’s break it down.
What Is Deep Research and How to Use It?
Normal Search mode on ChatGPT searches internet to provide real-time info. Whereas Deep Research, an extension of that, searches tens of sources and compiles a detailed report. So, if you want to learn about a specific topic in detail, conduct market research, or make big purchase decisions, you can use this feature and ChatGPT will generate a report. Basically, it summarizes everything on the internet related to that topic in a structured format. OpenAI uses o3 reasoning model to analyze and compile the report.
For example, if you ask a regular chatbot: “What are the health benefits of intermittent fasting?” it will simply summarize the surface level info. But if you use Deep Research, it will actually go online, scan research papers, articles, YouTube videos, Reddit threads, and blogs, then give you a structured facts and informations with sources linked at the bottom. In fact, it dwells down and searches deeper based on the information it found online.
OpenAI is giving 10 searches per month for ChatGPT Plus, Team, Enterprise, and Education subscribers. ChatGPT Pro users now get 120 searches per month, up from 100 searches.
Using Deep Research is easy. Here’s how you can try it:
- Open ChatGPT on web or mobile.
- Ensure you’re logged into a paid account (Pro, Plus, Team, Enterprise, or Edu).
- Type your research question and select Deep Research mode before clicking enter.
- ChatGPT will then ask a few questions to better understand your requirements. Since the process can take anywhere from 5 to 30 minutes, this step helps ensure the report is better aligned with what you need.
- Now ChatGPT will search all the sources, analyzes them and slowly compiles a detailed report.
That’s it! Instead of manually searching the web, opening multiple tabs, and piecing together information, which can take hours or even days to compile manually can now be done in few minutes.
Every AI Now Has Its Own Deep Research Model
Deep Research isn’t unique to OpenAI. Other AI models also have similar tools, but each works slightly differently.
Google’s Gemini Advanced was actually the first to introduce deep research. Similarly, it also searches 100s of sources to create a detailed report. However, Deep Research runs on Gemini 1.5 Pro which is a general purpose language model rather than a reasoning model. So even though it searches the same info, there is not much reasoning going into compiling that report. Compared to ChatGPT, Gemini keeps repeating the same info again and again and also we found various examples where it misunderstood the information.
However, it is worth noting that you get unlimited searches with Gemini Deep Research unlike ChatGPT.
Then we have Perplexity AI, which takes a different approach. Instead of spending time synthesizing a report, it prioritizes speed and pulls data from multiple sources instantly. When I tried “What is the latest research on sleep deprivation?” Perplexity instantly fetched results from WebMD, NIH, and Wikipedia, and compiled a quick report in just 3 minutes. This is far from ChatGPT’s 5 to 30 minutes of report.
And now there’s Grok (Elon Musk’s AI), which recently added its own Deep Research feature, powered by Grok 3. Like OpenAI’s Deep Research, Grok also uses a reasoning model, but early testing suggests it isn’t as thorough. When I tested “What are the effects of social media on teenage mental health?” Grok provided a decent summary but didn’t scan as many sources as ChatGPT’s model. It also felt more opinionated rather than deeply analytical.
In our experience overall, if you want a fast responses with cited sources, you can use Perplexity. But comparing to Gemini or Grok, ChatGPT does arguable better research most of the time and also compiles the report with it’s latest reasoning model.
