Exa Search API

aid: exa:search-api
name: Exa Search API
description: The Exa Search API provides neural web search, contents retrieval, find similar, and grounded
  answers for AI agents. Multiple speed modes range from fast 250ms searches to deep research lasting
  tens of seconds, and category indexes cover companies, people, news, and research papers.
humanURL: https://exa.ai/docs
baseURL: https://api.exa.ai
tags:
- Search
- Contents
- Find Similar
- Answer
- Highlights
- Summaries
- REST
properties:
- type: Documentation
  url: https://exa.ai/docs
- type: GettingStarted
  url: https://dashboard.exa.ai/onboarding
- type: SignUp
  url: https://dashboard.exa.ai
- type: APIReference
  url: https://exa.ai/docs/reference/search
- type: SDK
  url: https://github.com/exa-labs/exa-py
- type: SDK
  url: https://github.com/exa-labs/exa-js
- type: GitHubRepository
  url: https://github.com/exa-labs
- type: Pricing
  url: https://exa.ai/pricing
- type: Authentication
  url: https://dashboard.exa.ai/api-keys
features:
- name: Neural Search
  description: Embedding-based search across a purpose-built web index optimized for AI agents.
- name: Contents API
  description: Extract clean page contents and highlights from URLs returned by search.
- name: Find Similar
  description: Discover semantically similar pages given a seed URL.
- name: Answer API
  description: Generate grounded answers with citations using live search results.
- name: Speed Modes
  description: Fast, Auto, and Deep modes trade latency for comprehensiveness and reasoning depth.
- name: Category Indexes
  description: Specialized indexes for companies, people, news, research papers, and financials.
- name: Structured Output
  description: Extract structured JSON from results using an output schema across 70M+ companies.
- name: Token-Efficient Highlights
  description: Return condensed highlights to reduce LLM input tokens by up to 90 percent.
- name: Enterprise Controls
  description: SOC 2 Type II, Zero Data Retention option, and SSO for enterprise customers.
useCases:
- name: Agent Web Search
  description: Equip agents with a web search tool tuned for relevance and freshness.
- name: Deep Research
  description: Run multi-step research with citation-grounded answers and contents.
- name: Competitor and Market Intel
  description: Find similar companies, monitor news, and extract structured firmographics.
- name: RAG Over the Web
  description: Use search and contents endpoints as a web retrieval layer for RAG pipelines.
- name: Lead and People Search
  description: Query the People and Companies indexes for sales and recruiting workflows.
integrations:
- name: LangChain
- name: LlamaIndex
- name: CrewAI
- name: Vercel AI SDK
- name: OpenAI
- name: Anthropic
- name: Model Context Protocol
- name: LangGraph
- name: Pydantic AI
- name: Mastra
authentication:
- type: API Key
  description: Authenticate with an API key via the x-api-key header.