Case Study: Perplexity delivers factual, relevant search results with Anthropic’s Claude

Perplexity delivers factual and relevant answers with Claude

Perplexity, a conversational answer engine, sought to provide its users with the best possible AI-powered search experience by delivering factual and contextually relevant answers. To do this, the company needed to balance high performance and accuracy with financial sustainability for both its free and paid user tiers, which led them to partner with vendor Anthropic and utilize its Claude model family on the Amazon Bedrock platform.

By implementing a blend of Anthropic’s Claude models, Perplexity optimized its service for different user needs. The solution used the cost-effective Claude 3 Haiku for free users, Claude 3 Opus for 2x more accurate results for paid users, and Claude 3.5 Sonnet for 2x faster response times. This strategic deployment allowed Anthropic to help Perplexity enhance its platform's accuracy and performance, enabling it to offer a premium experience to paid users while maintaining a high-quality, financially sustainable service for everyone.

Perplexity

Aarash Heydari

Infrastructure Engineer