After launching the operator last week, Openai unveiled its next AI agent -deep research. The agent is designed to do detailed research using its chatbot, chatgpt. Customized for web browsing and python analysis operated by Openai O3, Deep Research has used arguments to browse text, image and PDFS on the Internet wisely and extensively, explain Openai.

In blog posts, it suggests, “Intensive research is designed for those who do intensive knowledge work in fields such as Finance, Science, Policy and Engineering and require completely and reliable research.”

“It is also useful for sensible shopkeepers, who are looking for hyper-reclosed recoses on shopping, which require careful research,” it is called.

Deep research of Openai: How to use

In Chatgpt, simply select “Deep Research” from the message musician and enter your query. Whether you need competitive analysis of streaming platforms or a personal report on the best computer bike, just tell the chat what you are doing. You can also attach files or spreadsheets for additional reference. Once you submit your request, it will appear to show the steps and sources used during a sidebar process.

Openai confirms that this AI agent can take anywhere from 5 to 30 minutes, based on the complexity of the investigation. After the research is complete, the user will receive a notification, and until then they are free to do other tasks. The final result will be given as a detailed report within the chat.

According to the company, while intensive research excel in providing wide, well -recorded reply with verified sources, GPT -4 is better suited for real -time, multimodal conversations. For complex, domain-specific inquiry that requires depth and accuracy, intensive research provides widespread investigation and benefit of citation-Restionable, correct for the production of actionable insight.

Explaining how this agent works, CEO Sam Altman posted an example on X.

Subsequently, Openai is also planning to add embedded images, data visualization and other analytical features to increase the clarity and reference of the report.

How is this different from the operator?

The major discrimination is that while the operator can be worked with a wide range of browser activities to be repeated, such as completing the form, ordering groceries, or even making memes, this new AI agent Can not do deep research.

Deep Research built on logic capabilities to bridge the difference, causing people to face people in work and everyday life.

However, both agents are an important step towards the broader target of developing AGI, which is pointing for a longer period.

It also comes at a time when a Chinese AI model, Dipsek, attracted a lot of attention for its advanced technology and cheap costs. Openai CEO Sam Altman also praised Deepsak, but promised a much better model.