OpenAI introduced a new tool “Deep Research” (Deep Research), which he developed to accelerate complex research processes and optimize access to information. This innovative feature offers users the opportunity to automate multi -step internet research and complete hours of analysis within minutes.

Deep Research aims to reach the most accurate data by creating and implementing a multi -step research plan. If necessary, it can revise the process by turning back and adapt to real -time information. Users can request more comprehensive and contextual research by offering various data types such as text, visuals, PDFs and tables. OpenAI says that this process may take 5 to 30 minutes and may include embedded visuals and graphics in the future.

What is Deep Research?

Deep Research is an artificial intelligence tool (Agent or Agent) that provides a comprehensive report to the user by scanning, analyzing and synthesizing large data stacks on the Internet. Strengthened by OpenAI’s O3 model, this system has been optimized, especially in web screening and data analysis. Artificial intelligence, according to the new information, can shape the research process and establish logical contexts between the sources can perform in -depth analysis.

For whom to Deep Research?

According to the company, Deep Research and Deep Research are designed for professionals working in areas that require intensive knowledge such as finance, science, engineering and politics. It also provides great convenience for users who want to make detailed product comparisons. For example, before you buy a car or electronic device, Deep Research can offer the most comprehensive analyzes for you without having to do hours of research to determine the best options. Probably it will be possible to collect the computer you want with this tool.

How does Deep Research work?

It is very simple to use the Deep Research:

1. Select “Deep Research” mode in Chatgpt and enter your research topic.
2. By providing additional files or data, you can make the system more accurately analyzed.
3. When the research begins, artificial intelligence scans, analyzes and synthesize hundreds of resources on the Internet.
4. A comprehensive report is prepared within 5 to 30 minutes and all resources are transparently specified.
5. With the updates to be added soon, visualized data, graphics and summary will be integrated into reports.

One more step to AGI

OpenAI describes the purpose of development of this technology as part of the vision of artificial general intelligence (AGI). The synthesis of information is seen as a critical step for the production of new information. Thanks to Deep Research, it will be easier to reach fast and reliable information in many areas from scientific research to decision -making processes in the business world.

How to use Deep Research?

Since Deep Research and deep research are constantly working on time, it requires much more calculation power than traditional question and answer models. The longer it takes to investigate a query, the more inferences I need to calculate.

Therefore, OpenAI opens its new vehicle to use 100 queries per month only for Pro subscribers for Pro subscribers. Plus and Team users will then be able to access Enterprise users. After that, free users will come.

Are there any limitations?

Ultimately, Deep Research is based on the “reasoning” talented O3. The “dreaming” problem, which is the sign of today’s artificial intelligence models, is also a risk factor in the new vehicle.

Deep Research has the capacity to collect and process information at a professional research analyst level, according to OpenAI. However, the company warns that this feature can sometimes produce incorrect information, that it may be difficult to distinguish reliable sources and rumors, and that it cannot correctly evaluate its answers correctly.

Deepseek made a difference to R1

OpenAI performed an artificial intelligence benchmark test at an expert level of Humanity’s Last Exam to test the accuracy of Deep Research. According to the results, the Deep Research model has reached the accuracy of 26.6 percent, while web scanning and Python vehicles are open, and the GPT-4O’s success has been increased by 3.3 %. Deepseek-R1, the most popular artificial intelligence of the recent period, was able to offer 9.4 percent accuracy. Despite these, the O3-Mini offered a 13 percent accuracy.