OpenAI Deep Research in ChatGPT // handle hard browsing and reasoning tasks across a range of domains
Are there open source alternatives? // Based on the best open models
Deep research was trained using end-to-end reinforcement learning on hard browsing and reasoning tasks across a range of domains. Through that training, it learned to plan and execute a multi-step trajectory to find the data it needs, backtracking and reacting to real-time information where necessary. The model is also able to browse over user uploaded files, plot and iterate on graphs using the python tool, embed both generated graphs and images from websites in its responses, and cite specific sentences or passages from its sources. As a result of this training, it reaches new highs on a number of public evaluations focused on real-world problems.
Mark from OpenAI introduces a new feature called Deep Research, designed to enhance AI’s ability to conduct multi-step research on the internet. Unlike traditional models, Deep Research allows extended reasoning, taking anywhere from 5 to 30 minutes to generate results, making it useful for complex tasks like market research, academic studies, and product recommendations.
Deep Research Capabilities
Can browse the web, open and analyze sources (including PDFs, images, and tables).
Produces comprehensive, cited reports similar to expert analyses.
Removes latency constraints, allowing longer, more thorough research.
Interacts with users by asking clarifying questions before conducting research.
Use Cases
Professional tasks: Market research, financial analysis, investment memos.
Academic research: Physics, biology, and other disciplines.
Consumer needs: Finding the best skis to buy, identifying TV shows from vague memories.
Technical Details
Powered by a fine-tuned O3 reasoning model trained via reinforcement learning.
Outperforms previous models on benchmarks like Humanity’s Last Exam and AGI capability evaluations.
Can use Python for calculations, generate visualizations, and embed images in responses.
Launch & Accessibility
Rolling out first to Pro users, then expanding to Plus, Team, Education, and Enterprise.
The demonstration showcased how Deep Research autonomously gathers, synthesizes, and presents information with citations, making it a significant step toward OpenAI’s vision for autonomous, knowledge-generating AI agents.
