The main goal of AI-focused companies is to develop AI agents. However, there are intermediate steps to take before reaching that point. OpenAI pointed the way with its Deep Research tool. Grok 3, xAI’s latest AI model, will offer a similar feature in the near future. Now, Perplexity is announcing its own approach to Deep Research.
What is AI Deep Research?
Deep Research is a feature that generates detailed reports on a specific topic according to your request. AI models use all their “analysis” and “understanding” capabilities to search for related information on the Internet, process it, and synthesize it. These types of searches take much longer than traditional prompts. However, Deep Research is not intended for the usual prompts but for more complex cases where you require a “white paper style” report.
The feature is not only useful for academics or researchers. It can also help you dive deeper into a specific topic by synthesizing all the information you might need. You can start the process with a single, detailed prompt instead of using multiple separate prompts.
Perplexity AI’s Deep Research tool is free; direct attack on OpenAI’s ChatGPT
That said, Perplexity becomes the first AI company to offer a Deep Research tool for free. There’s a daily usage limit if you don’t pay, though: five queries per day. Perplexity Pro subscribers ($20 monthly, $200 per year) will be able to use the full power of Deep Research for up to 500 queries per month. OpenAI, on the other hand, requires a paid ChatGPT Pro subscription ($200 monthly) to access Deep Research with a limit of 100 queries per month.
While their Deep Research underlying principles are the same, there are important differences between the Perplexity and Open AI approaches. OpenAI’s seems more complex, with higher processing requirements. Its Deep Research queries on ChatGPT are completed in around 20 minutes. On the other hand, Perplexity’s counterpart generates a response in just 2-4 minutes.
Great results in benchmarks
Despite the difference in price and time needed to get a response, it seems that Perplexity’s Deep Research is dangerously close to ChatGPT’s. The company revealed the results of its tool in Humanity’s Last Exam benchmark. This benchmark employs up to 3,000 prompts across various topics to challenge AI platforms to their limits. Perplexity’s tool scored 21.1% on accuracy, while ChatGPT’s Deep Research scored 26.6%.
Perplexity significantly outperforms other rivals, though. DeepSeek R1 and Gemini scored 8.6% and 7.2% on the same test, respectively. This is normal since neither DeepSeek nor Gemini have a Deep Research tool yet. Perplexity’s Deep Research tool is now available from the company’s website.