Perplexity’s Sonar: New API for AI Search Explained

Since its launch in 2022, Perplexity has often been considered one of the biggest threats to Google’s dominance in search. Having secured a spot in the list of best AI chatbots, it offers real-time, web-wide research capabilities that are not limited by training data.

Now, with its new Perplexity Sonar and Sonar Pro APIs, businesses can build their own generative search tools, ensuring factuality and authority. According to PwC’s survey, “73% of executives say they’ll use GenAI to make changes to their company’s business model.”

In this article, we’ve compiled everything you need to know about Perplexity’s new Sonar and Sonar Pro APIs.

Table of Contents Table of Contents

1. Key Takeaways
2. New Perplexity Sonar API Models for Custom Enterprise Solutions
3. Perplexity & DeepSeek: What Is Sonar Reasoning API?
4. Why Do Businesses Rush to Build Their Own GenAI Tools?
5. Perplexity’s Sonar API Real-Life Use Cases
6. Sonar APIs Future: a Whirlwind of AI Innovation & Risk
7. The Bottom Line: What Comes Next?
8. FAQs
9. References

Show Full Guide

Table of Contents

1. Key Takeaways
2. New Perplexity Sonar API Models for Custom Enterprise Solutions
3. Perplexity & DeepSeek: What Is Sonar Reasoning API?
Show Full Guide
4. Why Do Businesses Rush to Build Their Own GenAI Tools?
5. Perplexity’s Sonar API Real-Life Use Cases
6. Sonar APIs Future: a Whirlwind of AI Innovation & Risk
7. The Bottom Line: What Comes Next?
8. FAQs
9. References

New Perplexity Sonar API Models for Custom Enterprise Solutions

What Is Sonar API?

Sonar is the most affordable search API product on the market, Perplexity says.

On January 21, the company launched two tiers of API that developers can choose from depending on their needs – a cheaper and faster base version and a pricier Pro version that’s better for more complex questions.

Sonar vs. Sonar Pro Key Features

Perplexity & DeepSeek: What Is Sonar Reasoning API?

A week later, on January 29, 2025, Perplexity announced the launch of its new Sonar Reasoning API, powered by DeepSeek’s reasoning models.

Introducing Sonar Reasoning, our new API offering powered by DeepSeek's reasoning models.

Build products with chain-of-thought reasoning, plus real-time internet search and citations.

Sonar Reasoning is uncensored and hosted in US datacenters. As with Sonar and Sonar-Pro,… pic.twitter.com/Z5pKmDbnej

— Perplexity (@perplexity_ai) January 29, 2025

For context, according to DeepSeek’s research paper, R1 has been trained using reinforcement learning and chain-of-thought reasoning to increase the accuracy of the model’s responses.

The core selling point of R1, besides its open-source availability, is the low cost of its API. DeepSeek R1 API is available for $0.14 per million input tokens and $2.19 per million output tokens. In contrast, OpenAI’s o1 is available for $15 per million input tokens and $60 per 1 million output tokens.

Therefore, Sonar Reasoning API, powered by DeepSeek’s R1, will enable developers to build products with chain-of-thought reasoning, plus real-time internet search and citations.

Answering security concerns about DeepSeek’s model, Perplexity stated that the Sonar Reasoning API is hosted only in U.S. and EU data centers.

DeepSeek on Perplexity is hosted in 🇺🇸US/🇪🇺EU data centers – your data never leaves Western servers. The open source model is hosted completely independent of China. Your privacy and data security is our priority.

— Perplexity (@perplexity_ai) January 27, 2025

Beyond APIs, DeepSeek R1 is now available on Perplexity to support deep web research. A new Pro Search reasoning mode selector with a “transparent chain of thought into model’s reasoning” is available.

The company also increased the number of daily uses for both free and paid tiers to add more capacity.

Why Do Businesses Rush to Build Their Own GenAI Tools?

Uzi Dvir, CIO of WalkMe, highlighted the transformative impact of GenAI search capabilities. He told Techopedia:

“You no longer have to do the work of combing through stacks of results. What’s not as obvious is how AI-powered search will become integrated into employee workflows as businesses adopt powerful GenAI tools that account for a company’s entire data context.

 

“Imagine not even needing to prompt AI because it’s built to understand work needs and get people what they want faster, easier, and in a more digestible format. A transformed future of search is closely connected to future ways we work – and dramatic productivity gains.”

Can the new Perplexity API become this productivity booster for enterprises?

Traditional pre-trained AI models have limited capabilities. APIs with a real-time internet connection can significantly optimize their authority and factuality.

According to Perplexity, Sonar Pro is the best-performing model for factuality. The company stated:

“It combines the summarization power of LLMs with access to real-time information rather than relying on stored training data to answer questions. Sonar Pro leads this benchmark with an F-score of 0.858, while Sonar received an F-score of 0.773.”

Perplexity’s Sonar API Real-Life Use Cases

Numerous businesses have already used Perplexity Sonar in their operations, and thousands of developers in sales, finance, nonprofits, and advertising are adopting Sonar to power native search experiences.

Some of the most prominent examples include the following:

The use cases for Sonar API are actually limitless. For example, Sanchay, a final-year engineering student specializing in AI and data science, built a global app restrictions tracker using Perplexity’s Sonar API, R1’s reasoning, and Claude 3.5 for design just a few days after the API launch.

just shipped something cool using perplexity's sonar api to build a global status tracker, their search capabilities are awesome for pulling real-time data, built the entire frontend without any ui experience by using r1's reasoning and claude 3.5 to handle the design and vercel… pic.twitter.com/BIKHyaZc1P

— sanchay (@kernelkook) January 25, 2025

Sonar APIs Future: a Whirlwind of AI Innovation & Risk

AI-powered search engine Perplexity reportedly closed a $500 million funding round in December 2024, valuing the startup at $9 billion, which indicates a strong market appetite for the company’s solutions.

Aravind Srinivas, Co-founder and CEO of Perplexity, said in a LinkedIn post:

“The excitement around trying to make agents and vertical answer engines (legal, health, travel, academic, patents, etc) happen is pretty clear. We want to support it through Perplexity’s search-grounded LLM API: the most affordable generative search API in the market right now with real-time information, factual grounding, citations, and multi-step reasoning.”

Despite Sonar API’s promise of easy integration into any platform, significant customization may be necessary, tailored to specific industries and individual companies.

Commenting on what makes Sonar an exciting development, Sridevi Chodasani, Senior Full Stack Engineer at Cisco, said:

“Sonar feels like a glimpse into what’s next. It doesn’t just search, it thinks. And it’s already making search smarter, faster, and more human.”

Potential Challenges & Drawbacks of Perplexity AI

• Accuracy & misinformation: Perplexity provides grounded responses with citations. However, it can still generate incorrect or misleading information, especially if the sources contain inaccuracies.
• Limited context understanding: Depending on the context window, it may struggle with nuanced, multi-step reasoning or highly complex queries.

Context windows play a crucial role in determining the capabilities and limitations of large language models (LLMs).

According to Perplexity, “it’s crucial to consider the trade-offs associated with larger context windows. The computational costs of increasing context window sizes appear to grow quadratically, making it challenging to balance performance gains with resource requirements.”

• Dependency on sources: Perplexity’s answers’ quality depends on the sources’ reliability. The response will be compromised if the cited sources are biased, outdated, or low-quality.
• Lack of critical thinking: Although it might seem that Perplexity ‘thinks,’ it doesn’t. Unlike a human researcher, the tool can’t critically analyze or verify the credibility of sources.
• Privacy concerns: As with any AI tool that processes user queries, there may be concerns about data security and how user interactions are stored.

Not All That Glitters Is Gold

With such advanced capabilities for embedding AI into business operations – like the one provided by Perplexity’s Sonar API – no wonder organizations across industries are being “pulled into a whirlwind of AI innovation and risk,” said Gartner in its Top Strategic Predictions for 2025 report.

According to the forecast:

• By 2027, 70% of new employee contracts will include licensing and fair usage clauses for AI representations of their personas.
• By 2028, technological immersion will impact populations with digital addiction and social isolation, prompting 70% of organizations to implement anti-digital policies.
• 40% of large enterprises will deploy AI to manipulate and measure employee mood and behaviors.
• 30% of S&P companies will use GenAI labeling, such as “xxGPT,” to reshape their branding.
• 40% of CIOs will demand “guardian agents” be available to autonomously track, oversee, or contain the results of AI agent actions.

The Bottom Line: What Comes Next?

Although APIs like Sonar provide streamlined opportunities for the broader adoption of customized AI solutions for organizations at every scale, businesses should be innovative here.

Answering the question, “What comes after AI?” Gartner said:

“Any innovation is likely to arise because of AI. So, instead of trying to distinguish AI from other trends, we suggest clients embrace the combination.

“AI is part of a combinatorial disruption that will drive us to reexamine our assumptions and priorities about strategies, tactics, and innovation across the board. Thriving amid this combinatorial change means clients must build AI disciplines related to every mission-critical priority.”

AI is evolving as human use of the technology evolves.

FAQs