Introducing Exa 3.0Learn more
/search

The highest quality search API at every latency

Structured, token-efficient results for AI agents, drawn from the world's largest index of the public web and private information.

Try nowTalk to sales
Find recent AI agent product launches
Found 5 results

New Workspace Intelligence delivers unified, real-time understanding

workspace.google.com

New Workspace Intelligence delivers unified, real-time understanding to power agentic work.

What's new in Gemini Enterprise | Google Cloud Blog

cloud.google.com

Today, we are expanding Gemini Enterprise to meet this demand, transitioning AI from an isolated productivity tool into a secure, collaborative, autonomous engine for your business.

Google pushes Gemini toward 'agentic enterprise' with new platform

computerworld.com

Google has updated its Gemini Enterprise app, adding new ways for office workers to build, manage, and interact with AI agents.

Anthropic launches Claude Cowork for autonomous desktop tasks

anthropic.com

Claude Cowork autonomously handles desktop tasks, local files, and multi-step workflows for knowledge work.

OpenAI ships Operator agent with autonomous web browsing

openai.com

Operator is a new agent that can browse the web, fill forms, and complete multi-step tasks on your behalf.

Trusted by industry leaders and 900,000+ developers

Cursor
Cognition
HubSpot
Monday.com
Revolut
OpenRouter

Search the public web and proprietary data through one API

Give your agent fresh, comprehensive context through one search API. Exa brings together 100B+ indexed web pages, 3.5B+ pages crawled every day, and data on 70M+ companies, 211M+ places, 1B+ people, and more.

Read the docs

Accurate results and more complete coverage with semantic search

Exa searches by meaning, not just keywords, surfacing the most relevant content and the long tail that keyword search misses, even when a page never uses your exact words. Exa ranks #1 for quality on the Artificial Analysis Search Index, a leading independent benchmark for search providers.

Structured results that save tokens

Highlights pull out the passages that answer each query, and structured outputs return just the fields you need. Your agent uses 95% fewer input tokens on average without sacrificing accuracy, so every query costs less.

High-quality search for any workflow

Get the quality you need at the latency your workflow requires. Choose sub-second, cost-efficient single-step search for everyday queries or multi-step deep research for complex questions, all through one API.

Read about Exa 3.0

Outputs purpose-built for your agents

AI summary and structured outputs work out of the box.

The complete guide to AI workspaces
workspace.google.com
AI workspaces unify real-time data, automate routine tasks, and let cross-functional teams collaborate alongside autonomous agents that can plan, reason, and act on their behalf. Workspace Intelligence brings Gemini-powered context to every surface — Docs, Sheets, Slides, Mail, Drive, Calendar, Meet, and Chat — so agents can plan multi-step work, reason over your company's knowledge graph, and take action across the tools your team already uses every day. Built-in governance, observability, and granular access controls keep enterprise data safe while engineering and product teams ship new capabilities every single week. From drafting board-ready briefs and writing customer follow-ups to triaging inboxes, summarizing meeting recordings, generating dashboards from spreadsheet data, and orchestrating long-running agents that wait days for external dependencies, the platform spans the full breadth of modern knowledge work. Built on the same secure foundation that powers Google Workspace, with end-to-end encryption, data residency controls, audit logs, role-based access, BYOK options, and regional failover, it can scale from a single team pilot to a global, multi-region rollout with hundreds of thousands of seats without rearchitecting your access model.
Results

Standard search results. URLs and page contents.

New Workspace Intelligence delivers…
workspace.google.com

unified, real-time understanding to power agentic work

Exclusive: ClickUp endows its Brain assistant…
siliconangle.com

overhaul of its Brain assistant

Anthropic launches Claude Cowork…
anthropic.com

Claude Cowork autonomously handles desktop tasks

Highlights

Key passages extracted directly from source pages, ready to pass to your LLM.

"content": {
"Google": "Workspace AI"
"ClickUp": "Brain agents"
siliconangle.com
"OpenAI": "ChatGPT agent"
openai.com
"Gemini": "Agent tools"
}
Structured outputs

Extracted data fields, formatted for the exact data you need.

Google's Workspace Intelligence unifies real-time data across apps for agentic work [1]. Gemini Enterprise adds long-running agents and governance via Agent Platform [2][3]. Anthropic's Claude Cowork autonomously handles desktop tasks, local files, and multi-step workflows [4].
AI Summary

An LLM-summarized answer that can be streamed.

Highest accuracy at the lowest latency

Exa Search outperforms other web search APIs on accuracy benchmarks while staying the fastest in its tier.

Exa Lite69.6% / 454ms
Exa Auto67.8% / ~1.2s
Exa Deep80.0% / 4.3s avg
Parallel (Adv)46.0% / 2.7s
Parallel (Basic)41.0% / 1.3s
Brave43.0% / 547ms
Perplexity50.0% / 583ms
Accuracy / Latency

MultiLoko: Meta's multilingual benchmark of locally relevant questions across dozens of languages, testing search beyond English sources.

Teams building with Exa Search

CodeRabbit

“Exa consistently shows higher quality than other AI search providers we tested. What Exa is able to do in 1 search took our old provider multiple searches, taking 5 times as long.”

David Loker
David Loker
VP of AI, CodeRabbit
Read case study
HubSpot

“Exa is necessary for agents on HubSpot to supplement internal data with high quality, real-time people and company search. We found that other alternatives were slower, more expensive, and did not have comprehensive enrichment coverage.”

Praty Sharma
Praty Sharma
AI Lead, HubSpot
Read case study
OpenRouter

“OpenRouter has a laser focus on bringing a great developer experience to all language models, and Exa is the best way we've found for grounding AI in the real world in a model-agnostic way.”

Alex Atallah
Alex Atallah
CEO, OpenRouter
Read case study
Pricing

Search API for your agents

From $4/1k requests

Start free: your first team gets $10 in credits every month, plus a one-time $10 bonus when you complete onboarding. No payment method required.

Try it for freeTalk to sales
lite450ms

Lowest latency and cost

auto~1s

Adaptive search type that picks the best mode for each query. The best general-purpose default.

deep4–15s

Agentic research at 4–15s

Exa organizes data across billions of sources

Every datapoint is selected on source quality and freshness, based on the latest events from today

70M+ Businesses
Pricing pages
1B+ People Records
Job postings
Contact data
Product reviews
Latest News
Funding rounds
Financial market data
Crunchbase logo
Zoominfo logo
Similarweb logo
Harmonic logo
Financial Datasets logo
Baselayer logo
Definitive Healthcare logo
Fiber logo
Particle logo
Intellizence logo
Jinko logo
Faraday logo
Exa Connect
Earnings call transcripts
Investor portfolios
Recent hires
Conference speaker lists

Works with the tools you already use

Exa MCP logo

Exa MCP

Connect any MCP client

LangChain logo

LangChain

Build retrieval agents

LlamaIndex logo

LlamaIndex

Ground RAG pipelines

Vercel AI SDK logo

Vercel AI SDK

Add search to AI apps

OpenAI SDK logo

OpenAI SDK

Call Exa with the OpenAI client

CrewAI logo

CrewAI

Equip multi-agent crews

See all integrations (55)

Enterprise-grade security and compliance

Exa is built from the ground up to meet enterprise security and compliance needs.

Visit the Trust Center
SOC 2 Type II icon

SOC 2 Type II

Independently audited controls for security, availability, and confidentiality.

GDPR icon

GDPR

EU data-subject rights honored, with lawful-basis handling for personal data.

CCPA icon

CCPA

California consumer privacy rights, including access and deletion requests.

HIPAA icon

HIPAA

Safeguards for protected health information, with a BAA available.

Zero data retention icon

Zero data retention

Your queries and results are never stored or used for training.

See how to add ZDR
Enterprise agreements icon

Enterprise agreements

BAA and DPA documentation and custom data-security arrangements.

Common questions

The Exa Search API is a web search API built for AI agents. Send a natural-language query and get back ranked results with each page's URL, title, and publication date, plus its content as highlights, full text, or a summary. It searches Exa's own index of 100 billion documents.

A SERP API returns links and snippets from another search engine, and you fetch and clean each page yourself. Exa searches its own index by meaning, not just keywords, and returns page content in the same response. Text and highlights are included at no extra cost.

Lite returns in about 450ms at $4 per 1,000 requests, for the lowest latency and cost. Auto takes about 1 second at $7 per 1,000 and picks the best method for each query, which makes it the best general-purpose default. Deep runs agentic research in 4 to 15 seconds at $12 per 1,000, with structured outputs and grounded citations. For long-running research, list building and enrichment, use Exa Agent. See pricing for details.

Exa searches the public web plus dedicated data for news, companies and people, research papers, code, financial filings, legal opinions, and places. Set category to search a specific vertical.

Yes. Use includeDomains or excludeDomains to control sources, startPublishedDate and endPublishedDate to set a publication window, and userLocation to localize results. Each search returns up to 100 results.

Exa crawls 3.5 billion pages every day. For pages that change often, set contents.maxAgeHours to fetch a fresh copy of any page whose cached version is older than that limit.

Yes. Pass an outputSchema and Exa returns JSON in the shape you define, grounded in the pages it found. Use Deep when the output needs multi-step research across many sources. See the Deep Search guide.

Yes. Your first team gets $10 in free credits every month, plus a one-time $10 bonus when you complete onboarding in the dashboard. No payment method is required. The free plan also includes access to the Exa MCP server.

Exa is SOC 2 Type II certified. On Enterprise plans, zero data retention and HIPAA compliance can be enabled per team, and organizations get SSO. Review the SOC 2 report and DPA in the Trust Center.

Call the API directly or use the Python and JavaScript SDKs. Exa also works through the Exa MCP server and frameworks such as LangChain, LlamaIndex, and the Vercel AI SDK. See all integrations.