Perplexity has launched a new API service, Sonar. The new API allows developers and businesses to integrate its generative AI search tools into their applications. The API aims to provide lightweight, affordable, and customizable search capabilities, expanding Perplexity’s AI offerings beyond its app and website.

The company says Sonar API is designed for businesses seeking a cost-effective solution for AI-powered question-and-answer functionalities.

The versions of Perplexity’s API

Perplexity offers the Sonar API in two versions: Sonar API and Sonar Pro API. Both versions have different levels of AI search capabilities and customization needs.

Sonar API 

Perplexity’s Sonar API is the base version optimized for speed and affordability. It features flat-rate pricing and a lightweight AI model. Sonar API includes citations and source customization. The pricing structure for Sonar API includes, $5 per 1,000 searches, $1 per 750,000 input words, and $1 per 750,000 output words.

Sonar Pro API

Perplexity says Sonar Pro API is an advanced version designed for more complex, multi-step queries. The platform claims Sonar Pro API offers enhanced extensibility and double the number of citations compared to the Sonar API. It is priced at $5 per 1,000 searches, $3 per 750,000 input words, and $15 per 750,000 output words.

Expanding AI search beyond Perplexity's platform

With the launch of the Sonar API, Perplexity is expanding its AI search capabilities beyond its core app and website. 

The move positions the company to compete with major AI providers such as Google, OpenAI, and Anthropic, which have also introduced generative AI search solutions. The launch of the Sonar API also comes at a time when the AI industry has seen significant price reductions for API access. Many tech companies are lowering costs to attract more developers and businesses to their platforms. Perplexity claims sonar is the cheapest search API. 

Perplexity also claims Sonar Pro has outperformed leading AI models from its competitors in SimpleQA, a benchmark that measures factual accuracy in AI chatbot responses.

‍

‍

‍