Exa is a web search and content retrieval API built for developers creating AI agents, research tools, and applications that need current information from the web. Rather than optimizing search results for human clicks and advertisements, Exa is designed to give software systems relevant links, clean webpage content, concise excerpts, grounded answers, and structured research outputs through a programmable API.
Search Built for AI Applications
Exa provides several search modes so developers can balance latency, retrieval quality, and reasoning depth for each workflow. Its default auto mode selects an appropriate retrieval approach, while instant and fast modes support latency-sensitive experiences such as chatbots, voice agents, and autocomplete. Deep search modes handle more complex questions by expanding queries, gathering evidence across sources, and producing synthesized or structured results.
Searches can be focused on categories such as companies, people, research papers, news, personal sites, and financial reports. Developers can also control domains, publication dates, result counts, moderation, and other request parameters. This makes the API useful for broad web retrieval as well as targeted tasks such as company research, people discovery, documentation lookup, and current-news monitoring.
Content and Research Capabilities
- Search: Returns relevant webpages using Exa's search index and retrieval models, with modes suited to real-time and deeper agent workflows.
- Contents: Extracts webpage text, highlights, and summaries so applications can work with usable content instead of parsing pages themselves.
- Structured outputs: Lets developers define an output schema for extracting organized data from search results and synthesized research.
- Grounded answers: Produces answers backed by web sources and citations for applications that need traceable results.
- Monitors: Runs searches on a schedule and can notify applications about new events or changes through webhooks.
- Agent: Supports asynchronous deep research, list building, and enrichment tasks with structured outputs and citations.
Developer Experience and Integrations
Developers can call Exa through its HTTP API or use official JavaScript and Python SDKs. The documentation includes API references, quickstarts, tutorials, an API playground, and examples for common agent frameworks and model providers. Exa also offers an MCP server, making its search capabilities available to compatible AI assistants and development tools.
Content highlights are intended to reduce the amount of webpage text passed into language models by returning concise excerpts relevant to a query. Applications that need full context can request complete page text, while those extracting specific facts can request summaries or structured JSON. These options help teams control latency, token usage, and output format without building a separate crawling and extraction pipeline.
Use Cases and Pricing
Exa is suited to coding agents that need current documentation, chatbots that require web grounding, research assistants, news and event monitoring, company and people enrichment, and applications that need searchable web data. It can also support workflows that find similar pages, collect source material, or turn web research into structured records.
Exa offers a free tier with up to 1,000 requests per month. Paid usage is priced by endpoint, including separate rates for Search, Deep Search, Contents, Monitors, Answer, and Agent operations. Enterprise options add higher limits, custom datasets and indexes, security features, service agreements, onboarding, and volume discounts.
