Our own visit statistics are always on. Google Analytics, which helps us understand how people use our public pages (not your signed-in workspace), is enabled only with your permission. See our Privacy policy.

Exa

Exa is a search API company built for AI agents. You describe what you want in plain language. Exa returns ranked web results and page content that you can pass straight into a model's context.

Visit Exa(opens in a new tab)

What Exa offers

Capabilities and prices follow the Exa website, its pricing page and the developer docs, read on October 24, 2026. Exa prices by request tier, and its free credit and rate limits can change, so recheck them before you plan volume.

Exa capabilityHow Exa explains itPrice listed by Exa
Search (Lite, Auto and Deep)Natural-language queries, 10 results by default and up to 100; filters by domain, path and locationFrom $4, $7 and $12 per 1,000 requests
Page contentsHighlights most relevant to the query, full text or summariesFrom $1 per 1,000 pages
Structured outputAdd an output schema and Exa turns results into text or JSON, with citationsBilled with the search type
Answer and MonitorsCited answers; ongoing monitoring of a topicFrom $5 and $15 per 1,000 requests
MCP serverWorks with Claude, Cursor, VS Code and others; Exa describes it as free, open source and key-freeFree

The Exa website also states:

  • The index covers the web, people, companies and scholarly works. The developer docs explain what is inside.
  • The pricing page lists $10 in free credits each month and 10 queries per second. Pay-as-you-go raises this to 25 per second and adds SOC 2 Type II.
  • Enterprise plans add zero data retention, HIPAA, a custom contract and dedicated support.

Who it is for

  • Developers of AI agents and retrieval apps who want models to read current pages, docs and code, not only training data.
  • Sales and investment research teams who look up companies, people and similar sites by description.
  • Researchers who search papers and long articles by meaning.
  • Claude and Cursor users who want to give an assistant web search through MCP.

Prospect research: Exa finds the companies, EveryInfra checks their social accounts

B2B sales and investment teams often start a list from one sentence, such as "D2C pet food brands in Southeast Asia". Exa's index covers web pages, people and companies, so a natural-language query is a good way to find them. To decide which ones to contact, you then need to see what they actually do on social platforms.

  1. Build the list. Run the sentence through Exa search with structured output, so results come back as JSON with company name, website and country, plus citations.
  2. Check company accounts. For each company, use the EveryInfra LinkedIn company page and jobs endpoints to get headcount, follower count and the roles it is hiring for now.
  3. Check brand activity. Use the Instagram profile and posts endpoints for followers, posting frequency and engagement on recent posts. For brands aimed at China, also look up the Xiaohongshu account.
  4. Score. Rank the list by open roles plus engagement rate, and let sales contact the 20 most active first.

If a company has no LinkedIn or Instagram account to find, EveryInfra returns an empty result and does not charge. EveryInfra also has its own 17 web search and reading tools (EverySearch); for plain web search, run the same queries through both and compare results and price.

Try one query on Exa

The Exa pricing page lists $10 of free credit per month, so you can test a few searches with your ideal customer description and judge the list. Its MCP server is free and needs no API key. Exa publishes its current prices and policies at exa.ai; use that page when you budget.