Exa — Search engine and web data API built for AI agents (search, contents/crawl, answer, monitors, research agent).. Pricing model: Usage-based pay-as-you-go with free monthly credits; Enterprise custom, Usage-based. Value metric: Per 1,000 requests by search depth; agents per compute unit and effort level.
At a glance
| Company | Exa |
|---|---|
| What it does | Search engine and web data API built for AI agents (search, contents/crawl, answer, monitors, research agent). |
| Headquarters | San Francisco, California, USA |
| Pricing model | Usage-based pay-as-you-go with free monthly credits; Enterprise custom, Usage-based |
| Value metric | Per 1,000 requests by search depth; agents per compute unit and effort level |
| Free tier | $10 credits/month plus $10 onboarding bonus (~$120/year); startup and education grants up to $1,000. |
| Pricing page | exa.ai/pricing |
Plans and prices
Prices as published on Exa's pricing page, checked 27 October 2026. They change often; check the live page before you rely on them.
| Plan | Price | Billing | What you get |
|---|---|---|---|
| Lite Search | $4 per 1K requests | usage | Up to 10 results; +$1 per 1K per extra result |
| Auto Search | $7 per 1K | usage | |
| Deep Search | $12 per 1K | usage | |
| Contents | $1 per 1K pages | usage | |
| Answer | $5 per 1K | usage | |
| Monitors | $15 per 1K | usage | |
| Agent | $0.10 per Agent Compute Unit; effort levels $0.012 (minimal) to $1.00 (x-high) per request | usage | Search tool call $0.005; email enrichment $0.02; phone $0.07 |
Usage component. All usage-based; enterprise pricing available.
How credits/tokens work. Effort-level pricing lets the caller choose a per-request price ceiling.
What changed in 2025–2026
2026: agent pricing with effort levels and Agent Compute Units added.
Growth milestones
Exa appears in our analysis of 2026's $100M ARR class. Figures are company-reported unless stated.
- 2026-01-12: ARR reported at $12M. (source)
- 2026-05: $250M Series C led by a16z (Nvidia, YC participating) at ~$1.95B (TechCrunch) / $2.2B (ARR Club) valuation. (source)
- 2026-10-14: Passed $100M annualised revenue, ~10x per year for three years; 200,000+ teams, 2B requests/month (CEO Will Bryk). (source)
How it grows
Developer-first PLG: free monthly credits, MCP server and integrations into agent frameworks; startup grants; becomes default search tool inside AI products, so usage grows with customers.
What you can steal
Pricing
Letting the customer pick an effort level sets a known price per request even for agent work. A menu of fixed price points is easier to buy than an open-ended meter.
usage-based pricing
Recurring free monthly credits (not just a one-off trial) keep small users active until their usage turns into a bill.
low-budget growth
Credit grants for startups and educators seed future large customers cheaply.
moats
Switching: once embedded as a tool inside customers' agents, replacing you means re-tuning their product.
Sources
Get both books free
The Subscription Playbook (ebook + audiobook) and Feed a Starving Crowd, instantly. In Australia? Claim the free hardcover.
Robert Coorey, MBA
Robert is the author of The Subscription Playbook and Feed a Starving Crowd and a Director at BPAG Accountants in Sydney. More about Robert.