Perplexity Today (February 14) announced the launch of Deep Research Perplexity is designed to help users dramatically save time and perform in-depth research and analysis on their behalf. When a user makes a Deep Research query, Perplexity performs dozens of searches, reads hundreds of sources, and analyzes the material in an inferential manner to generate a comprehensive research report on its own.Deep Research specializes in a wide range of expert-level tasks, including finance, marketing, and product research, and has achieved significant scores on Humanity's Last Exam benchmark. 's Last Exam benchmark test.Comments Off on Perplexity's new Deep Research release shows its ambition in the AI research space by directly targeting deep research tasks that traditionally take human experts hours to complete, the review noted. deep research claims to have performed well in several benchmarks, signaling that Perplexity A major upgrade to the product.
In line with Perplexity's mission to make powerful research tools available to everyone, Deep Research is now available to all users for free, with unlimited queries for Pro subscribers and a limited number of queries per day for free users. Deep Research is available today on the Web and will soon be available on iOS, Android and Mac platforms. (Be sure to update your app to the latest version.)COMMENTARY: The review suggests that the free strategy is a common tactic for Perplexity, which has helped it to rapidly expand its user base. In particular, Deep Research is also available to free users as a premium feature, which remains attractive despite the limitations on the number of times it can be used.
Users can start their experience by visiting perplexity.ai and selecting "Deep Research" in the mode selector of the search box.
Deep Research is a similar program that originated in STORM , the well-known commercial products known to have been released so far are:
OpenAI Launches DeepResearch , Intelligence for Deep Research Using o3 Models
Deep Lake proposes a Deep Research program built on private multimodal data.
Jina Inferential Deep Search Tool DeepSearch API is now live!
Working Principle
Perplexity's excellence in question answering is well documented, and the launch of Deep Research takes question answering to a whole new level, accomplishing in 1-2 minutes what would take a human expert hours to do. Here's how Deep Research works:
- A Study in Reasoning - The Perplexity Deep Research model has powerful search and coding capabilities, enabling iterative search and document reading, and based on this, reasoning about where to go next. Deep Research is able to continuously optimize the research plan as the understanding of the subject area grows. This is similar to the way humans research new topics, deepening their understanding as they continue to explore.Comment: The review notes that the concept of "Reasoning Research" emphasizes the intelligence and autonomy of Deep Research. It is no longer limited to simple keyword matching searches, but is more like a human expert, exploring and understanding the problem in progressively deeper ways.
- Report Writing - Once the source material has been fully evaluated, the Intelligence integrates all the findings into a clear and comprehensive report.
- Export & Share - Users can export the final report to PDF or document format, or convert it to a Perplexity page for easy sharing with colleagues or friends.
Application Scenarios for Deep Research
Perplexity has developed Deep Research to empower users to perform expert analysis across a range of complex disciplines, with particular expertise in generating specialized results in areas such as finance, marketing and technology. Deep Research also plays an important role in personal applications such as health advice, product research and travel planning. Here are some examples of how you can use Deep Research:
financial
marketing
skill
the present situation
well-being
biographies
travel
Humanity's Last Exam Benchmarking
Perplexity Deep Research achieved an accuracy of 20.5% on the Humanity's Last Exam benchmark, significantly higher than the Gemini Thinking, o3-mini, o1, DeepSeek-R1 and many other leading models.Humanity's Last Exam is a comprehensive benchmark for AI systems, containing more than 3,000 questions covering more than 100 disciplines from math and science to history and literature.Comments: Comments suggest that the name "Humanity's Last Exam" is quite attractive, and that Perplexity Deep Research's results in this benchmark are compelling and directly comparable to a number of large models, including Google's Gemini. Perplexity Deep Research's results in this benchmark are compelling and directly comparable to several large-scale models, including Google Gemini, demonstrating Perplexity's confidence in the performance of its Deep Research.
SimpleQA Benchmarking
Perplexity Deep Research in SimpleQA The benchmark achieved an accuracy of 93.9% (SimpleQA is a knowledge base of thousands of questions used to test AI's factual quizzing capabilities), far exceeding the performance of other leading models.COMMENTARY: The review notes that Deep Research achieved an accuracy of 93.9% on the SimpleQA benchmark, an excellent figure that demonstrates Perplexity Deep Research's superior performance in factual quizzing, which is critical for research tasks where information accuracy is critical.
Operational efficiency statistics
Not only did Perplexity Deep Research top industry benchmarks, it also completed most research tasks in less than 3 minutes - and Perplexity is working to increase its speed even further in the future.Comment: The review identified high performance and fast response as the two key elements that users are most concerned about. If Deep Research can realistically accomplish complex research tasks in minutes, this will significantly improve research efficiency and could have a profound impact on the traditional research paradigm.
Visit perplexity.com today, select "Deep Research" in the search box, and experience for yourself the highly effective research capabilities of Perplexity Deep Research.