Case Study: Perplexity delivers more accurate, faster search results with Claude

Perplexity delivers factual and relevant answers with Claude

Perplexity, a conversational AI-powered answer engine, sought to provide its users with the best possible search experience by delivering factual, relevant, and comprehensive answers. The challenge was to balance high performance and accuracy with cost-effectiveness across both its free and paid user tiers. To accomplish this, the company turned to vendor Anthropic and its Claude model family.

The solution was implemented using Anthropic's Claude models via Amazon Bedrock. Perplexity uses Claude 3 Haiku for its fast, cost-effective free tier and offers the more advanced Claude 3 Opus and Claude 3.5 Sonnet to its paid users. This multi-model approach from Anthropic yielded significant results, including achieving response times two times faster with Claude 3.5 Sonnet and delivering answers that are two times more accurate with Claude 3 Opus, all while optimizing for operational costs.

Perplexity

Aarash Heydari

Infrastructure Engineer