OpenAI reveals its most powerful tool yet, designed for “deep research”
- ChatGPT deep research turns hours of human time into minutes of compute time
- It’s built on the o3 model and promises more context than o1
- Available for ChatGPT Pro now, Plus and Team next
OpenAI has lifted the wraps off a new ‘deep research’ tool which it says is designed for in-depth, complex research tasks within ChatGPT.
Described as a, “new agentic capability that conducts multi-step research on the internet for complex tasks,” deep research promises to do hours of human work in tens of compute minutes.
Much like a human research worker, it will “find, analyze, and synthesize hundreds of online sources to create a comprehensive report at the level of a research analyst,” OpenAI said in its announcement.
OpenAI creates its own researcher
The tool is built on an upcoming version of the OpenAI o3 model, which is specifically optimized for optimized for web browsing and data analysis, and is capable of handling text, images and PDF files that can all be accessed via the internet.
Although the mind first arrives at scientific research, keen to demonstrate deep research’s usefulness, OpenAI says it’s particularly strong at helping consumers make decisions based on “hyper-personalized recommendations,” including some of the biggest decisions we’re likely to make such as buying cars, appliances and furniture.
Given the extent to which deep research trawls the internet, the documented output includes citations to help users validate information sources and ensure accuracy.
OpenAI says the o3-based deep research also addresses the shortage of context and other information that o1 currently struggles with – that model is better suited to coding, maths and other technical domains.
It’s important to note that, while many ChatGPT users have become accustomed to near-instantaneous responses, deep research could take 5-30 minutes to complete a task. Users will receive a notification once the research is complete, so they won’t have to hang around to wait for a response.
The response comes in the form of a report within the chat interface, but embedded images, data visualizations and other analytic outputs are in the pipeline and could arrive in a matter of weeks.
Deep research is already available for Pro users paying $200 per month, but Plus ($20/month) and Team ($25/month) subscribers are also set to get access to the powerful upgrade.