In a dramatic turn in the AI saga, OpenAI has launched Deep Research to reclaim its title as the leader in AI. This announcement comes just after China’s DeepSeek made waves in the AI world.
Deep Research: The New Contender
OpenAI designed Deep Research to autonomously plan and execute a wide range of research tasks across the internet. In this model, users provide a prompt through ChatGPT, and the bot searches the internet for various types of information—such as text, images, or PDFs—to generate a detailed report. This process significantly reduces the time and effort humans would typically spend on report writing.
DeepSeek’s Meteoric Rise
Meanwhile, DeepSeek has stirred excitement with its efficient and affordable AI model, DeepSeek R1. The company developed this model at a fraction of the cost of its Western counterparts while still matching the capability benchmarks of state-of-the-art models like OpenAI’s GPT-4. Its open-source development and rapid rise have disrupted the tech industry, directly challenging the established AI hierarchy. I found the comments under the Deep Research introduction video quite funny, take a look!
How Deep Research Works
See ChatGPT’s ‘deep research’ for your query and any attached files. The AI processes all information available to it to extract sources and report in detail within five to thirty minutes. There’s a sidebar showing its progress through to completion and notifies you when the work is done. Reports will include illustrations, data visualizations, and analysis to clear up any obscurities.
Unlike GPT-4o, which is excellent for conversations in a real-time mode, deep research is used for complex domain-related questions with good documentation for its answers.
Deep research attempts to plan, provide backward reasoning, search through files, and generates graphs for insights from reinforcement learning. It garnered 26.6% in the ‘Humanity’s Last Exam’ and set the new state-of-the-art on GAIA, mastering in reasoning, multimodal fluency, and web browsing.
The AI Arms Race Intensifies
OpenAI sees this as its next strategic move to counter DeepSeek’s disruptive entry into the AI sphere. By enhancing ChatGPT’s research output, OpenAI aims to stay competitive and meet the rapidly growing demand for specialized AI tools. DeepSeek’s rise has exposed increasing conflicts of interest worldwide in AI development.
Furthermore, as AI titans clash, they are transforming the world of AI at lightning speed. OpenAI’s Deep Research and DeepSeek’s R1 model mark major breakthroughs in artificial intelligence, pushing the boundaries of what was previously thought possible. As these developments unfold, one thing remains certain: the AI race is far from over. On a slightly different note we put ChatGPT and DeepSeek to the ultimate April fool’s prank test, check it out to find out who won.
Update July 2025 – Key Developments in Deep Research
- In April 2025, OpenAI launched a light weight version of Deep Research, which is actually powered by the o4-mini model. How does this version activate? It automatically does when users finish their quota on the full Deep Research model.
- In June 2025, OpenAI introduced Deep Research in its API suite. In order to support o3 and o4-mini models, webhooks, as well as web search.
- The latest development that took place was on July 17, 2025, where OpenAI integrated Deep Research into the new ChatGPT Agent. Which literally allowed it to autonomously use the tool for complex, multi-step tasks and in-depth analysis, as reported by TechCrunch.
