Perplexity Search API
by Perplexity
High-throughput, AI-optimized search API returning raw ranked web units for RAG and agent workflows
See Perplexity Search API docs and SDK
Summary
Perplexity launched the Search API in late September 2025 to provide programmatic access to its web index and retrieval pipeline. Unlike its Sonar API, the Search API returns raw, ranked web snippets / units (optimized for machine consumption) rather than synthesized answers.
Features
- High-throughput access to hundreds of billions of web documents
- Returns ranked, machine-friendly web units for grounding LLMs
- SDKs for Python and TypeScript, plus a search_evals framework for evaluation
- Real-time freshness: tens of thousands of index updates per second
Superpowers
Designed for teams building RAG systems, agents, and search-first AI tools that need up-to-date web signals and transparent grounding sources.
Pricing
- Example pricing reported: $5 / 1,000 requests (confirm current pricing on Perplexity docs)
Known limitations & notes
- Returns raw units not synthesized answers — integrate downstream synthesis or ranking as needed
Sources / notes:
- Perplexity product announcements and SDK docs.