Perplexity has joined the growing list of AI companies launching in-depth research tools, with a new feature introduced on Friday. Following similar releases from Google with its Gemini AI platform in December and OpenAI’s launch of its own research agent earlier this month, all three companies have introduced a feature called “Deep Research.” The aim is to provide more thorough responses with accurate citations, targeting professional use cases, and offering a more in-depth experience than what is typically seen with consumer chatbots.
Perplexity’s blog post announcing the Deep Research feature describes its capabilities, noting that it excels at a variety of expert-level tasks, including those in finance, marketing, and product research.
Currently available on the web, the Deep Research feature will soon be integrated into Perplexity’s apps for Mac, iOS, and Android. To use it, users can simply select “Deep Research” from a drop-down menu when submitting a query. The tool will then generate a detailed report, which can be exported as a PDF or shared as a Perplexity Page.
Perplexity’s Deep Research tool works by iteratively searching, reading documents, and refining its research plan as it learns more about the topic, mimicking the way a human might conduct research on a new subject.
In a demonstration of its capabilities, Perplexity highlighted its performance on Humanity’s Last Exam, a test that evaluates AI models with expert-level questions across a wide range of academic fields. Perplexity’s Deep Research scored 21.1% on the test, outperforming many other models, such as Google’s Gemini Thinking (6.2%), Grok-2 (3.8%), and OpenAI’s GPT-4o (3.3%), though it fell short of OpenAI’s Deep Research score of 26.6%.
While OpenAI’s Deep Research tool requires a $200-per-month Pro subscription, Perplexity’s version is available for free, with non-subscribers getting a limited number of queries each day. Paying subscribers have unlimited access. Perplexity’s tool is notably faster, typically completing tasks in under three minutes, compared to OpenAI’s Deep Research, which can take between 5 to 30 minutes to complete similar tasks.
When comparing the tools, Perplexity’s Deep Research is praised for its speed and accessibility, making it ideal for casual researchers. OpenAI’s Deep Research, on the other hand, is recognized for its analytical depth, especially for enterprise-level applications. Google’s offering stands out for seamlessly integrating with productivity ecosystems.
As these tools continue to evolve, their impact on everyday and professional research remains uncertain. A recent article in The Economist raised concerns about OpenAI’s Deep Research, pointing out potential drawbacks such as limited creativity in data interpretation, a reliance on easily accessible sources, and the risk that heavy reliance on AI tools could reduce opportunities for individuals to generate their own ideas. These concerns may also apply to other Deep Research tools as they become more popular.
