The final response from deep research includes a detailed report with suggestions and citations from relevant sources.

The deep research tool performs multi-step research analysis and citation collection directly within the GitHub repository.

The ability to click on sources for validation helps users trust the data returned by AI agents, addressing concerns about accuracy and reliability.

The sharing feature of Deep Research makes it easy to distribute generated reports, which is a valuable addition for users.

The clarification flow in Deep Research allows the model to ask users for more information, enhancing the interaction and improving the accuracy of the responses. To get the best response from the research model, the prompt needs to be well-specified and detailed, as users typically don't provide all necessary information in the first prompt.

The model that powers Deep Research is a fine-tuned version of the O3 model, specifically trained on challenging browsing and reasoning tasks, enhancing its online search and analysis capabilities.

High-quality training data is crucial for the success of Deep Research, as the quality of the input data significantly influences the model's output quality.

We're excited to continue developing deep research, improving its reliability, and integrating it into our main reasoning models.

In the medical field, Deep Research can help doctors quickly find relevant literature and clinical trials, saving them time and improving patient care.