Perplexity AI launches Deep Research tool to take on OpenAI, Google

Perplexity AI

In a move to challenge AI giants like OpenAI and Google, Perplexity AI has introduced its new Deep Research tool. The tool — as the name suggests — can help users with expert-level analysis on a range of complex topics, from finance and marketing to health and travel planning.

Perplexity's Deep Research tool is designed to save users hours of time by conducting in-depth research and analysis autonomously. According to CEO Aravind Srinivas, the tool can complete most tasks in under three minutes, making it significantly faster than its competitors. The tool performs dozens of searches, reads hundreds of sources, and synthesises the information into a clear and comprehensive report.

“Excited to introduce the Perplexity Deep Research Agent: available for free to all users. Paid users only need to pay $20/mo to access an expert level researcher on any topic for 500 daily queries, and need to wait less than three minutes for getting a full research report,” said Srinivas in a post on X.

The tool is currently available on the Perplexity website and will soon be rolled out to iOS, Android, and Mac versions of the app.

Perplexity claims that its Deep Research tool can provide expert-level analysis across various domains, including finance, marketing, technology, health, product research, and travel planning. The tool iteratively searches, reads documents, and reasons about what to do next, refining its research plan as it progresses. This process is similar to how a human researcher might approach a topic, said Perplexity.

“We built Deep Research to empower everyone to conduct expert-level analysis across a range of complex subject matters. Deep Research excels at creating work artifacts in domains including finance, marketing, and technology, and is equally useful as a personal consultant in areas such as health, product research, and travel planning,” said the AI firm.

While Perplexity's Deep Research tool falls behind ChatGPT in accuracy rankings on the Humanity's Last Exam, it outperforms other models like Gemini Thinking, o3-Mini, Grok 2, and Claude 3.5 Sonnet.