OpenAI rolls out new ChatGPT agent for 'Deep Research'

Deep Research is designed to support individuals in knowledge-intensive fields such as engineering, finance, literature, and biology.

OpenAI has introduced a groundbreaking tool called Deep Research, showcased on a YouTube livestream on Sunday, shortly after being presented to lawmakers, policymakers, and officials in Washington.

OpenAI Founder Sam Altman announced the launch on microblogging site X, stating, "Today we launch deep research, our next agent....this is like a superpower; experts on demand..it can go use the internet, do complex research and reasoning, and give you back a report...it is really good, and can do tasks that would take hours/days and cost hundreds of dollars."

today we launch deep research, our next agent.this is like a superpower; experts on demand!it can go use the internet, do complex research and reasoning, and give you back a report.it is really good, and can do tasks that would take hours/days and cost hundreds of dollars.
Sam Altman (@sama) February 3, 2025

The new mode for ChatGPT allows the AI to perform intricate, multi-step research across the internet, marking a significant stride towards what OpenAI calls "agentic AI"—technology designed to autonomously carry out human-like tasks.

One of the key features highlighted by OpenAI is the model's lack of latency constraints, enabling real-time data access from the internet. However, this feature means responses may take longer to generate. "We believe it's essential for our models to undertake autonomous tasks for extended periods without supervision," OpenAI developers explained during the live stream, positioning this development as a step towards achieving artificial general intelligence—AI capable of understanding and executing any intellectual task at a human level.

Deep Research is designed to support individuals in knowledge-intensive fields such as engineering, finance, literature, and biology. Initially, the model will be accessible to OpenAI’s $200-a-month Pro users, with plans to extend availability to the $20-a-month Plus tier pending safety evaluations within the next month.

Due to its high computational demands, Deep Research tasks may take anywhere from five to 30 minutes based on input complexity. OpenAI assured that the model will provide outputs with clear citations and transparent reasoning, though it acknowledged that the tool remains susceptible to occasional inaccuracies, or "hallucinations," albeit at a reduced rate compared to other ChatGPT models.

This launch follows the recent release of another OpenAI tool capable of performing online tasks like grocery shopping and restaurant reservations. Deep Research advances this capability further by gathering and synthesising information from the internet into concise, well-documented reports.