Discover the New DeepSeek R1 Model in Perplexity
For those who have been contemplating the use of DeepSeek, an emerging AI application from China, but hesitated due to issues surrounding privacy and censorship, there’s a noteworthy update. A modified version of the DeepSeek R1 model has now been integrated into Perplexity.
Perplexity has established itself as a platform that offers users various AI choices for prompt responses, with the addition of DeepSeek R1 as the latest offering. Perplexity’s co-founder, Aravind Srinivas, assures that any censorship impositions have been removed, allowing discussions on sensitive topics like Tiananmen Square. Furthermore, all user data, including prompts and answers, is kept in the U.S.

Credit: DailyHackly
This integration is feasible due to the open-source nature of DeepSeek models, enabling anyone to utilize and modify them without cost. While tech-savvy individuals might choose to adjust these AI models independently, it’s generally advisable to allow experts—like those in the Perplexity team—to handle such modifications. More details on this can be found here.
Perplexity has specifically included the DeepSeek R1 reasoning model, distinguishing it from the broader V3 model. Users can engage with these models regardless of their subscription level: free users receive five queries a day, while those who subscribe for $20 monthly can access up to 500 queries daily.

Credit: DailyHackly
If you’re eager to explore the rapid processing and capabilities of DeepSeek without the anxieties pertaining to data sharing with the Chinese government or being restricted by local censorship, utilizing it through Perplexity is an excellent choice. While your data is managed by Perplexity in compliance with their privacy policy, it is securely housed in the U.S.
Moreover, several other AI providers are also embracing this model: Microsoft is integrating DeepSeek R1 into Windows, particularly within the Copilot feature, while it has also been deployed on You.com AI platform, where it complements models from various companies, including Anthropic, Meta, Grok, and OpenAI.
How to Use DeepSeek Within Perplexity
Accessing Perplexity can be done via the website, or through Android and iOS applications. On the website, prompts can be entered using the main text entry field on the Home tab or by selecting New Thread on the left side. To engage the DeepSeek R1 and receive its answers, switch on the Pro toggle and select Reasoning with R1 from the dropdown options.
For mobile users, the process is equally straightforward. Simply tap into the Ask anything box on the central tab of Perplexity (represented by the magnifying glass icon in the lower left), activate the Pro option, and choose Reasoning with R1 to commence submitting inquiries.

Credit: DailyHackly
As the DeepSeek R1 operates as a reasoning model, it elaborates on how it interprets your prompts while generating responses. Each answer includes citations from relevant websites, and you will find options to copy, share, or regenerate the provided answers if desired.
An examination of the Perplexity version of DeepSeek revealed that it handles potentially controversial topics, such as Taiwan, without restriction, yielding an uninhibited flow of information. In contrast, directly asking the original DeepSeek for the same inquiries returned a polite message saying, “Sorry, that’s beyond my current scope. Let’s talk about something else.”

Credit: DailyHackly
The reasoning capabilities of DeepSeek R1 appear to be quite advanced. The transparency of showing how answers are derived, paired with the consulted sources, enhances trust in the AI’s responses. Users can review this information both during and after the response generation process.
Previous comparisons highlight how Perplexity stands up against competitors like Gemini and ChatGPT, combining elements of generative AI with web search and offering an array of AI model options. Additionally, Perplexity has recently introduced a Perplexity Assistant feature for Android users, facilitating the use of AI instead of traditional assistants like Gemini or Bixby.