Perplexity AI Inc. today introduced an application programming interface that will enable developers to embed search features into their applications.
The API is available in two editions: Sonar and Sonar Pro. Both use the company’s Sonar series of large language models to process search queries. The API’s release comes two months after Perplexity reportedly raised $500 million in funding at a $9 billion valuation.
Founded in 2022, Perplexity provides an AI-powered search engine of the same name. When users enter a query, the service doesn’t return a list of web page links but rather answers the question directly in natural language. Each prompt response includes citations to the webpages from which Perplexity gathered the retrieved information.
The company generates revenue with a paid version, Perplexity Pro, that offers more features. A built-in code interpreter enables Perplexity Pro users to not only generate code but also run it. Meanwhile, a tool called Shop like a Pro makes it possible to make e-commerce purchases in one click through the search engine’s interface.
The API that Perplexity debuted today will provide the company with another way of monetizing its LLMs.
The API’s base edition, Sonar, is optimized for fast query response times and cost-efficiency. It’s powered by a Perplexity-developed LLM of the same name with a context window of 127,000 tokens. This means users can include up to 127,000 tokens’ worth of data in each prompt.
Perplexity also offers a more capable version of its API, Sonar Pro, that is better at processing complex queries. It’s powered by an LLM called Sonar Pro that has a context window of 200,000 tokens. It includes twice as many citations in prompt responses as the standard Sonar API to help users verify the accuracy of the retrieved information.
Perplexity has not shared technical details about the LLMs that power its API. However, the company did disclose that one of the earlier models in the same LLM series is a customized version of Llama 3.1 70B.
On launch, Perplexity provides a handful of customization options for API customers. Developers can customize the data sources on which the API draws to answer search queries. Users may also adjust the underlying LLMs’ Top P and presence penalty settings, which help reduce nonsensical output and avoid duplicate content in prompt responses, respectively.
The Sonar API is priced at $5 per 1,000 searches, plus $1 for every 750,000 words the LLM ingests or generates as output. Sonar Pro is billed at $3 per 750,000 words of input, while the same amount of output data costs $15.
Perplexity says that thousands of developers are already using its API in software projects. The company claims that one customer, startup Copy.ai Inc., is using the service to save several hours of work per week for customers of its sales platform.
Image: Perplexity
Your vote of support is important to us and it helps us keep the content FREE.
One click below supports our mission to provide free, deep, and relevant content.
Join our community on YouTube
Join the community that includes more than 15,000 #CubeAlumni experts, including Amazon.com CEO Andy Jassy, Dell Technologies founder and CEO Michael Dell, Intel CEO Pat Gelsinger, and many more luminaries and experts.
THANK YOU
