Perplexity launches ‘Deep Research’ tool to rival OpenAI & Google

Perplexity, the AI-powered search platform, has introduced a new feature called 'Deep Research,' joining the race to provide in-depth, citation-backed answers for professional use cases. The tool, announced on Friday, follows similar offerings from OpenAI and Google, both of which have also named their features Deep Research.  

The feature aims to deliver detailed reports on complex topics, catering to fields such as finance, marketing, and product research. Users can access 'Deep Research' via a drop-down menu on Perplexity’s web platform, with plans to expand it to its Mac, iOS, and Android apps soon. Reports generated by the tool can be exported as PDFs or shared as Perplexity Pages.  

Perplexity claims the tool operates “similar to how a human might research a new topic,” iteratively searching, reading documents, and refining its approach as it gathers more information. The company highlighted its performance on 'Humanity’s Last Exam', an AI benchmarking test, where it scored 21.1%, outperforming models like Google’s Gemini Thinking (6.2%) and OpenAI’s GPT-4o (3.3%), though falling short of OpenAI’s 'Deep Research' (26.6%).  

Unlike OpenAI’s 'Deep Research', which is currently limited to a $200-per-month subscription, Perplexity’s version is free for all users, albeit with a daily query limit for non-subscribers. Paying subscribers enjoy unlimited access. Perplexity also boasts faster processing times, completing most tasks in under three minutes compared to OpenAI’s 5 to 30 minutes.  

While these tools promise to revolutionise professional research, concerns remain. A report recently noted potential limitations, including reliance on easily accessible sources and reduced opportunities for human creativity when outsourcing research to AI.  

As competition heats up, Perplexity positions itself as a faster, more accessible option for casual researchers, while OpenAI focuses on analytical depth for enterprise applications, and Google integrates its tools into existing productivity ecosystems. The long-term impact of these tools on research practices remains to be seen.