Following the launch of its o3 mini model, **OpenAI** has introduced a novel ChatGPT feature aimed at facilitating thorough research. Aptly named “Deep Research,” this tool is intended to serve as a virtual research assistant for handling intricate assignments.
Built on an upcoming version of OpenAI’s reasoning model, o3, Deep Research can execute multi-step internet research. As per OpenAI, the tool is designed for users involved in demanding knowledge work across sectors such as finance, science, policy, and engineering. Nevertheless, it is also beneficial for everyday tasks, like evaluating choices while purchasing a new vehicle.
### A Recognizable Idea in the AI Competition
If the notion of Deep Research rings a bell, it’s because **Google Gemini** rolled out a comparable feature—also called “Deep Research”—in early December. This development highlights the ongoing rivalry between OpenAI and Google as both strive for supremacy in the AI industry. Both organizations are currently concentrating on enhancing advanced reasoning skills and constructing agents capable of undertaking tasks autonomously on behalf of users.
Complicating the competitive scenario is China’s **DeepSeek**, which recently attracted attention with its budget-friendly R1 model. The launch of R1 has shaken up the tech sector, unsettling stock markets and raising alarms among AI creators due to its cost-effectiveness and potential.
### OpenAI’s Reaction to Rivals
In spite of the excitement surrounding DeepSeek’s R1, OpenAI has been intensifying its initiatives through a series of new offerings. These include the o3 mini model, the “Operator” agent for internet browsing and task automation, and now the Deep Research feature. OpenAI CEO Sam Altman recognized DeepSeek’s accomplishments in a recent statement on X (formerly Twitter), labeling R1 “an impressive model” while expressing his belief in OpenAI’s capacity to produce superior products. “We will obviously deliver much better models,” Altman remarked, noting that the rise of new competitors is “invigorating.”
Altman further reiterated OpenAI’s dedication to its research strategy and its objective of achieving artificial general intelligence (AGI). He stressed the significance of computational power in meeting these ambitions, pointing to the billions of dollars raised by OpenAI to support its lofty goals.
### Functionality of Deep Research
Unlike models such as GPT-4o, which excel in real-time dialogue tasks, Deep Research is tailored for exhaustive, time-consuming projects. Tasks may take between five to 30 minutes to finish, enabling users to step away from their devices while the tool operates. Upon completion of the research, users are notified with a detailed report directly in the chat.
The reports produced by Deep Research encompass detailed outputs, summaries of its reasoning methodology, and citations complete with links to its sources. This level of transparency equips users with insights into how the tool reached its conclusions.
### Access and Availability
Given the substantial computational power necessary for each query, OpenAI is rolling out Deep Research in phases. Initially, it will be accessible to **ChatGPT Pro** users, who will have 100 queries available each month. The feature will subsequently extend to Plus, Team, and Enterprise users. Currently, Deep Research is exclusively available via the web, but OpenAI intends to introduce mobile and desktop app support by the end of February. However, users in the UK and EU will face a longer wait, as OpenAI is still in the process of making the feature available in those areas.
### Concluding Remarks
As OpenAI continues to hone its offerings, Deep Research signifies another progress in the development of AI tools aimed at assisting with complex, knowledge-driven assignments. With rivals such as Google and DeepSeek pushing the frontiers of AI innovation, the competition to lead the next phase of AI development is intensifying.