OpenAI launches deep research agent for $200 a month

OpenAI has introduced a new agent for ChatGPT called "deep research," designed to perform comprehensive online research and generate detailed reports.

This feature enables users to conduct multi-step research tasks similar to the work of a research analyst.

According to OpenAI's blog post on February 2, the deep research agent is tailored for professionals in fields such as finance, science, policy, and engineering, as well as for consumers looking to make informed purchases.

The outputs from the deep research agent can take between 5 to 30 minutes to complete and include clear citations and summaries of the research process.

OpenAI emphasised that this tool aims to enhance the efficiency of knowledge work by providing thorough and reliable information.

The deep research agent utilises the latest OpenAI o3 model, optimised for web browsing and data analysis.

In a test known as Humanity’s Last Exam, which includes 3,000 expert-level questions across various topics, the deep research agent achieved an accuracy rate of 26.6%.

This performance significantly outstripped other models, including DeepSeek-R-1, which scored 9.4%, and GPT-4o at 3.3%.

Despite its capabilities, OpenAI cautioned that the deep research agent may still produce inaccuracies or "hallucinate" facts in its responses.

It may also struggle to differentiate between authoritative information and rumors.

The deep research feature is currently available to Pro users at a subscription cost of $200 per month, with a limit of 100 queries monthly.

Plans are in place to expand access to Plus and Team users in the future.

This launch follows a competitive landscape in AI, as Google also announced a similar feature for its Gemini model.