Build AI assistants with all the knowledge in the world

Add Exa as a tool your model can call. When a question needs current information, your app sends it to Exa, and Exa returns ranked pages with their URLs, dates, and the passages that answer the question, so the model can write its reply with citations for each source. See the Exa Search API quickstart or the Anthropic and OpenAI tool calling guides.

Connect your agent to Exa Search, which searches the live web instead of relying on training data. One search call covers news, sports scores, weather forecasts, local businesses, and financial markets, each result with its source URL. For topics a user follows, Exa Monitors rerun a search on a schedule and send only new results to your webhook.

Create an Exa Monitor with a search query, a schedule, and a webhook URL. On each run, Exa reruns the search, deduplicates against results it already sent, and posts new updates to your webhook as a text summary or as JSON in your pre-defined schema. Monitors cost $15 per 1,000 requests.

Yes. With Highlights turned on, Exa returns only the passages from each page that answer the query, each with its source URL. In Exa's evaluations, Dynamic Highlights cut tokens by an average of 95% compared with full pages.

Yes. Microsoft retired the Bing Search APIs in August 2025. Exa does the same job for agents and applications that rely on search. Your app sends a query and gets ranked web results, cited passages, or a direct answer the model can quote. Exa Search starts at $4 per 1,000 requests.

Exa is pay-as-you-go, with no subscription or minimum spend. Exa Search starts at $4 per 1,000 requests with the instant search type, and each request includes up to 10 results. Page contents cost $1 per 1,000 pages. New accounts get $20 in free credits, about 2,800 searches, and the free tier adds $10 every month. Rates for every endpoint cover deep search, Monitors, and Exa Agent, and Enterprise plans add volume discounts and custom rate limits.

The free plan allows 10 search queries per second, and pay-as-you-go plans go up to 25. Enterprise plans set custom rate limits for search and for Exa Agent concurrency.

All three return web results an LLM can cite. Exa is the stronger choice for searching people, companies, and code, and for agent loops where latency is critical. Tavily fits teams that want full page text bundled into every result and high-volume general lookups. Brave costs $5 per 1,000 requests versus Exa's $7 and scores higher on citation precision, while Exa scores higher on groundedness.

Yes. Exa is SOC 2 Type II audited and honors GDPR and CCPA data-subject rights. Enterprise plans offer zero data retention, HIPAA with a BAA, SSO and SCIM, and a custom MSA and DPA. To access audit reports, request them in the Trust Center or talk to our team.

Firefox uses Exa to power Smart Window on desktop and Quick Answers on iOS. Firefox chose Exa after evaluating multilingual quality, latency, safety, and privacy.