Perplexity API Platform

Production-grade primitives. Fully composable.

Perplexity runs on these same primitives, so every improvement to the product improves the API too.

Agent API: supported models, web search, URL fetching and code execution in one agent.

Explore Agent API
  • Top frontier models, plus selected open-source models.
  • Multi-step reasoning with built-in web access.
  • Adjustable tools and skills: web search, URL fetch, image search, people search, finance search and sandbox.
  • Structured, source-cited responses.
  • Effort presets, from fast to extra high, updated as evaluations show improvements.

Search API

Web search built for agents.

See OpenRouter's search benchmarks →Explore Search API

Orchestrator API

Every supported open-weight model, hosted by Perplexity.

Request access

Embeddings API

Embeddings for search and retrieval.

Explore Embeddings API

Decisions API

Answers as probabilities, not prose.

Explore Decisions API

Customer stories

Already in production

From consumer agents to public-sector retrieval, teams are shipping on Perplexity APIs today.

Ship your first request in five minutes.

Open the playground, then move to production with Python and TypeScript libraries plus framework integrations.

View the SDKs

Frequently asked questions

  • Start with Agent API for a full agent, Search API for raw retrieval or Orchestrator API for model choice.