Exa Search API
Exa Search API provides web search for AI agents with configurable search modes, webpage content, highlights, structured outputs, and AI generated summaries.
About Exa Search API
Exa Search API is a web search API designed for AI agents and applications that need current information from the web. Developers can send search queries to the /search endpoint and receive relevant web results, including URLs and page content that can be supplied to downstream applications and models.
The API offers multiple search modes for different latency and retrieval requirements. Exa documents Instant Search, Fast Search, Auto Search, Deep Search, and Deep Reasoning Search. Auto selects a search mode based on the query, while Deep Search supports more involved retrieval and can return structured output based on a developer supplied schema.
Search responses can include webpage text, highlights containing relevant passages extracted from source pages, structured data, and AI generated summaries. Exa provides Python and JavaScript libraries alongside direct REST access, with REST requests authenticated using a Bearer API key.
Key Features
- Web Search Endpoint: Send POST requests to
/searchto retrieve relevant web results for applications and AI agents. - Multiple Search Modes: Supports Instant, Fast, Auto, Deep, and Deep Reasoning search options for different latency and retrieval requirements.
- Webpage Content: Search requests can return webpage content together with search results.
- Highlights: Extract relevant passages directly from source pages for use as compact context in AI applications.
- Structured Outputs: Deep Search can return extracted information according to a developer supplied output schema.
- AI Summary: Search can produce an LLM generated summary, with streaming supported for the summarized answer.
Pricing
-
Instant Search: $4 per 1,000 requests Base price includes up to 10 results per request.
-
Fast / Auto Search: $7 per 1,000 requests Base price includes up to 10 results per request.
-
Deep Search: $12 per 1,000 requests Base price includes up to 10 results per request.
-
Deep Reasoning Search: $15 per 1,000 requests Base price includes up to 10 results per request.
-
Additional Search Results: $1 per 1,000 requests for each additional result above 10 Applies to Instant, Fast / Auto, Deep, and Deep Reasoning Search.
-
AI Page Summaries: $1 per 1,000 pages Available as an additional content processing component.
-
Free: $10 in credits per month plus a $10 onboarding bonus No payment method is required. Includes access to all endpoints and up to 10 Search queries per second.
-
Pay As You Go: Usage based Supports up to 25 Search queries per second with payment by card or bank account.
-
Enterprise: Contact the sales team for pricing information. Includes custom Search QPS, enterprise SLAs, custom agreements, zero data retention, and support for up to 1,000 results per search.
Pricing last updated: October 5, 2026 at 12:00 AM
Use Cases
- Give AI agents access to current web search results
- Retrieve current documentation, repositories, and technical information for coding agents
- Ground chatbots and AI assistants with web information
- Retrieve and structure web information for research and enrichment workflows
Pros & Cons
Pros:
- Provides multiple search modes covering low latency retrieval through deeper search workflows
- Can return webpage content, extracted highlights, structured outputs, and AI generated summaries
- Provides Python and JavaScript libraries plus direct REST API access
Cons:
- Free access is limited to 10 Search queries per second and Pay As You Go supports up to 25 Search queries per second
- Custom Search QPS and support for up to 1,000 results per search are Enterprise features
Integrations
Python, JavaScript, cURL, OpenAI, Claude, Vercel AI SDK, CrewAI, Haystack, Mastra, LangChain, LlamaIndex, IBM watsonx, Pydantic AI, Instructor, Langfuse, Letta, NVIDIA NeMo Agent Toolkit, n8n, Make, Google Sheets, Groq, Together AI
FAQ
What does Exa Search API do?
Exa Search API lets developers search the web and retrieve relevant URLs, webpage content, highlights, structured outputs, and AI generated summaries for agents and applications.
How do developers access Exa Search API?
Developers can use Exa's Python or JavaScript libraries or send REST POST requests directly to the /search endpoint using Bearer API key authentication.
What search modes does Exa provide?
Exa documents Instant, Fast, Auto, Deep, and Deep Reasoning Search. These modes provide different latency and retrieval characteristics.
Can Exa return structured data from search?
Yes. Deep Search supports an output schema that can be used to return structured information for specified fields.
Does Exa Search API provide content for LLM applications?
Yes. Search can return webpage content and extracted highlights, and it can also produce AI generated summaries.
Last edited
October 5, 2026 at 8:47 AM by Venkatraman Chandrasekaran
