Perplexity API Platform

Perplexity API Platform is a developer platform for building AI applications and agents that need access to language models, real-time web information, retrieval capabilities, and model-routing services through a unified set of APIs. It is intended for software teams embedding these capabilities in their own products, workflows, assistants, and internal tools—not for employees using the Perplexity Enterprise application directly.

Its principal component, the Agent API, provides a unified interface for building LLM-powered applications. Developers can select models from multiple providers, configure reasoning and token budgets, preserve conversation context, and add tools such as web search and URL content retrieval. The API returns web-grounded responses with inline citations when search is enabled, enabling use cases such as research assistants, customer-facing answer experiences, monitoring workflows, and domain-specific agents. It supports Perplexity’s SDKs for Python and TypeScript and accepts an OpenAI-compatible Responses-style integration path, reducing migration work for applications already built around that format.

The Search API is available when an application needs search results rather than a generated answer. It returns real-time, ranked web results as structured data, with controls for domains, languages, regions, multiple queries, and extracted content. Developers can use these results in their own ranking, retrieval, citation, or agent workflows.

The platform also provides an Embeddings API for semantic search, retrieval-augmented generation (RAG), recommendation, and other similarity-based applications. It offers standard embeddings for independent texts and contextualized embeddings for document chunks that benefit from shared document context.

A separate Router API, currently in private preview, provides unified access to open-weight models hosted by Perplexity. It supports OpenAI Chat Completions and Responses formats as well as Anthropic Messages, routes requests to healthy deployments, and is designed to let developers change models or migrate compatible applications with minimal integration changes. Because it remains in preview, it should be understood as an emerging component of the platform rather than its primary market identity.

The API Platform includes developer administration features such as API key management, projects and billing, usage analytics, rate-limit tiers, official SDKs, and pay-as-you-go billing. Model, tool, search, and embedding usage are metered separately according to the relevant API and service.