Following the recent launch of o3-mini, OpenAI has introduced a new product—Deep Research, an AI assistant specifically designed for automating complex online multi-step research tasks. Its functionality bears a clear resemblance to Google’s search engine and DeepSeek’s research assistant. Users simply input prompts, and Deep Research can quickly search, analyze, and consolidate hundreds of information sources on the internet, ultimately producing a comprehensive report at a professional level. This capability reduces research tasks that would typically take humans several hours to complete down to around 30 minutes.
Deep Research targets users across various professional fields, including finance, science, policy, and engineering. Users can utilize it to interpret financial reports, analyze experimental data, and even study legal cases. When faced with choices for high-value items like cars and appliances, Deep Research can also provide personalized consumer recommendations. This tool relies on an optimized version of the o3 model and employs end-to-end reinforcement learning technology, enabling cross-modal searches and the ability to understand and analyze text, images, and PDF files.
Using Deep Research is very straightforward. Users just need to select the "Deep Research" mode in the ChatGPT interface, input their research requirements, and they can even upload reference materials. The entire research process will be displayed in real-time on the sidebar, and users will receive a notification once the research is complete. Future updates will also enable Deep Research to support more charts and visual content to enhance user experience.
However, the launch of Deep Research has not been without controversy, especially due to its high similarity to DeepSeek, which has led to mockery on social media. Despite this, OpenAI has shown impressive performance in technical benchmark tests with Deep Research, including results from HLE, GAIA, and Expert-Level Tasks, all exceeding those of DeepSeek. These tests not only assess AI performance across various disciplines but also evaluate its ability to handle real-world problems.
Although Deep Research demonstrates powerful capabilities, OpenAI acknowledges that it still has some limitations, such as issues with information accuracy and report formatting. As usage increases, OpenAI plans to continuously optimize the model in an effort to address these issues. Users are eagerly anticipating Deep Research, especially regarding the additional features it may introduce in the future.
