How to use Deepseek R1 on Perplexity

The world of artificial intelligence has turned upside down with the arrival of the model DeepSeek R1, a revolution that has not only left experts and companies speechless, but has also generated controversy and some suspicion in the technology sector. The integration of DeepSeek R1 into Perplexity AI and other platforms has marked a turning point. for those looking to get the most out of Artificial Intelligence tools, both for individual users and businesses.

This article breaks down everything you need to know about DeepSeek R1 on Perplexity: from how it works and what it offers in practice, to its advantages over other models, key aspects regarding censorship and security, and comparisons with other solutions on the market.

What is DeepSeek R1 and why has it revolutionized artificial intelligence?

 

DeepSeek R1 is an advanced AI model developed by the Chinese company High-Flyer, led by Liang Wenfeng. It is a model designed for advanced reasoning tasks, especially excelling in mathematics, programming, logical analysis, and natural language. Mixture of Experts (MoE) architecture With 671 billion parameters and reinforced training, it is at the same level as – and in many cases above – other cutting-edge AIs such as OpenAI, Google Gemini y Microsoft Copilot.

One of its strong points is efficiency, since although its size is colossal, it only activates a part of its parameters in each query, which allows a speed and a much lower cost compared to direct competitors. For example, DeepSeek R1 can cost up to 27 times less per million tokens compared to models like OpenAI's o1.

In terms of reviews and benchmarks, DeepSeek R1 has been enthusiastically received by companies such as Nvidia and OpenAI, who have recognized its potential and its contribution to the advancement of AI in complex tasks.

DeepSeek R1 integration with Perplexity AI: a leap forward

 

Perplexity AI, one of the most innovative AI-based search engines today, has taken the step of Integrate DeepSeek R1 into your web and app servicesThis means that users can access the power of DeepSeek directly from the Perplexity interface, taking advantage of its reasoning capabilities and ease of use.

The implementation of DeepSeek R1 at Perplexity stands out for several reasons: Key aspects:

• Uncensored results: Perplexity has adapted the model to remove the original restrictions imposed by the Chinese government on political or sensitive topics, allowing for greater freedom of consultation and transparency of information.
• Privacy and secure storage: The data processed by DeepSeek R1 at Perplexity is stored on servers in the United States and Europe, allaying concerns about control or access from China.
• Open source and availability: DeepSeek R1 is an open source model, making it easy to integrate and use by both developers and businesses.

How to use DeepSeek R1 in Perplexity: steps and recommendations

The first step is Access the Perplexity platform, available both on the web and in the iOS appIf you don't have an account yet, create one easily. Once logged in, follow these steps to take advantage of DeepSeek R1:

1. Locate the search box and select the “Reasoning with R1” option (it may appear as simply “reasoning” in some versions).
2. You can optionally customize the fonts from which to obtain the information, which adds extra control over the results.
3. Enter your query and let DeepSeek R1 put all its reasoning skills to work. If you need to view the sources of information, there's always a dedicated tab called "Sources."
4. Imaging: A unique feature allows you to generate images from the corresponding icon, making it even more versatile for presentations or visual documentation.

Important: For free users, access to DeepSeek R1 is typically limited to 3-5 queries per day. However, those who opt for a Perplexity Pro subscription can perform up to 500 searches per day, making it ideal for researchers, students, or intensive business applications.

Differentiating features compared to other AI models

• Cost and efficiency: DeepSeek R1 is much cheaper to operate than other AIs at the same level, offering an unbeatable price per million tokens.
• Transparent and sourced answers: One of the greatest advantages of using DeepSeek R1 in Perplexity is that it shows the intermediate steps of its reasoning and cites the sources used, something that is not common in the competition.
• Long query support: It can handle contexts of up to 128.000 tokens, allowing you to parse long documents or complex text strings.
• Uncensored output: The Perplexity R1 1776 version is trained not to block responses on sensitive topics, unlike the original DeepSeek R1, which can refuse to address certain issues.
• Advanced Integrations: It allows you to interact with multiple apps, book restaurants, debug code, or schedule reminders all from the same interface, streamlining both your personal and professional work.

The Perplexity R1 1776 model: the DeepSeek without filters

One of the most notable advances has been the creation of the R1 1776 variant, exclusive to Perplexity, which completely removes the censorship and restrictions imposed on the original DeepSeek R1 model. This "R1 1776" has been trained on over 40.000 multilingual messages and 300 sensitive topics to ensure open responses to historical, political, and human rights-related questions. The number "1776" symbolically refers to freedom, harkening back to the year of the American Declaration of Independence.

Thanks to this effort, users can now Obtain accurate information and reasoning without the model refusing to talk about previously forbidden topicsInternal studies by Perplexity and specialized media indicate that more than 1000 queries on previously blocked topics have been successfully answered, with a level of accuracy far superior to other supposedly "uncensored" AIs.

Advantages for businesses and advanced users

The availability of the DeepSeek R1 model on high-level platforms such as Azure AI Foundry, GitHub, and Google's Vertex AI has facilitated its adoption by large companiesIn fact, Microsoft and Google have confirmed the integration of R1 among their available models, adding it to a catalog of more than 1.800 open and commercial models.

The most notable advantages for the business and professional environment are:

• Scalability and speed: DeepSeek R1 can be quickly and securely integrated into enterprise workflows using trusted infrastructures like those from Microsoft and Google.
• Comprehensive security assessments: Microsoft, for example, has subjected the model to red team testing and security reviews to mitigate any potential risks.
• Integration flexibility: It can be easily deployed from repositories like Hugging Face, making it easy to access even for independent developers.

Pricing plans and access to DeepSeek R1 on Perplexity

DeepSeek R1 is Available for all types of users, but the experience varies depending on the plan selectedFor those who want to quickly explore or integrate its capabilities, the free version offers up to 3-5 queries per day.

• Free version: Allows up to 3-5 searches per day using the R1 model in Perplexity. Enough for those who only need specific queries.
• Perplexity Pro: For $20 a month (and free for customers of certain operators like Movistar for a year), the range expands to 500 daily queries and exclusive features, ideal for those who require intensive or professional use.

Take advantage of Perplexity Pro free annual offer It can be especially interesting if you are a customer of operators that have reached agreements with the platform, something that is becoming popular in Spain and Latin America.

Future Perspective: Is DeepSeek R1 the New Standard in AI Reasoning?

The emergence of DeepSeek R1 and, above all, His uncensored arrival at Perplexity has changed the rules of the game.Thanks to its cost-effectiveness, reasoning power, and thematic freedom, it is expected to become one of the benchmark models in the coming years, both at home and in work and educational settings.

The trend indicates that More and more platforms will integrate open models like DeepSeek R1, and the user will gain freedom, transparency and control over information.Even so, the debate about security and sensitive content will continue, so we will see constant improvements to achieve the balance between innovation and responsibility.

It's clear that we're at an exciting stage in the history of artificial intelligence. DeepSeek R1, at Perplexity, offers a powerful and flexible tool that enables students, professionals, and the curious to obtain detailed answers, conduct in-depth research, or solve technical challenges without the limitations imposed by previous models.