Exa

Search the live web semantically, then attach an email for two cents

Exa is a web search and data API built for AI systems rather than people, selling semantic search over its own index at 7 dollars per thousand requests, page content retrieval, agentic deep research, and a Websets product that runs a described prospecting query, verifies each result against your criteria, and enriches it with contact details at 0.02 dollars for an email and 0.07 for a phone number, all pay-as-you-go with no subscription or minimum spend.

Visit website

Overview

Exa is not a contact database and it is important to say so first. It is a search engine with its own index of the live web, built by Will Bryk and Jeff Wang in 2021 on the premise that search designed for humans is the wrong shape for software. Where a conventional B2B data vendor sells you a pre-built table of people and companies, Exa sells you the ability to ask the web a question in natural language and get structured, verified results back. The contact enrichment sits at the end of that pipeline rather than being the product.

That difference is the whole reason a small business would look at it. Every other vendor in this category answers the question of how to reach a person you can already name, or lets you filter a fixed database by title and headcount. Exa answers the harder question of who exists at all. A Websets query such as agtech companies in the United States that raised a Series A is not something Apollo, Skrapp, or Hunter can express, because it is a research task rather than a filter. Exa runs it, checks each candidate against the criteria you set, then enriches what survives.

The pricing is the most transparent in this batch and possibly in the category. Search is 7 dollars per thousand requests. Page contents are 1 dollar per thousand. Answers with citations are 5 dollars per thousand. Deep search runs 12 to 15 dollars per thousand. Monitors, which watch the web on a schedule for new matches, are 15 per thousand. Contact enrichment is 0.02 dollars for an email address and 0.07 for a phone number. New accounts get 20 dollars in credits, roughly 2,800 searches, plus 10 dollars a month thereafter, with no subscription and no minimum spend. Nothing is quoted, nothing is gated, and nothing expires into a monthly reset.

The company has raised heavily on the strength of that positioning: 85 million dollars in a Series B led by Benchmark in September 2025 at a 700 million dollar valuation, then 250 million in a Series C led by Andreessen Horowitz in May 2026 at 2.2 billion. That is a lot of capital behind an API, and it cuts both ways for a small buyer. Exa is not going to disappear, and it is also not primarily building for a five-person sales team. If you want a finished interface, buy something else. If you want the underlying capability at commodity prices, this is a genuinely different kind of tool.

Best for

Technical teams and operators building list discovery or research into their own workflow, anyone whose prospecting question is a description rather than a filter, and buyers who want commodity pay-as-you-go pricing with no subscription and no expiring credits.

Not the right fit for

Non-technical salespeople who want a finished prospecting interface; Exa has a dashboard but the product is an API, and a team that wants to open a tool each morning will be happier with Apollo, Skrapp, or CUFinder.
Anyone who needs cheap bulk email lookups; at 0.02 dollars an email Exa is more than twice Tomba's rate and it is not competing on that metric.
Buyers who need phone numbers at volume; 0.07 dollars a mobile is reasonable but Exa is a search company that enriches, not a phone data specialist, and it publishes no match rate for either data type.
Teams needing a CRM, a sequencer, lead lists, or any workflow layer; Exa produces structured data and hands it over, and everything downstream is your problem.
Organisations that need contractual data accuracy guarantees or documented B2B contact data provenance; Exa's compliance framing is that of a search company, not a contact data vendor, and it says little about where enrichment data originates.

How it works

1
The base operation is a semantic search against Exa's own index. You describe what you are looking for in natural language rather than in keywords, and Exa returns matching pages, charged at 7 dollars per thousand requests for the first ten results and 1 dollar per thousand for results beyond that.
2
Contents retrieval pulls the full text, highlights, or an AI summary from URLs you already have, at 1 dollar per thousand pages, with summaries adding another 1 dollar per thousand. This is the cheap path when you know where to look and only need what is there.
3
Websets is the prospecting product. You describe a set, such as companies of a certain type meeting certain conditions, and Exa searches asynchronously across many candidates, verifies each one against rules you define, and only then runs your enrichments on the survivors, so you do not pay to enrich results that fail your criteria.
4
Enrichments extract specific data points you specify from each verified result, including contact details, with email enrichment at 0.02 dollars and phone enrichment at 0.07. Because verification runs first, the enrichment spend lands only on rows you actually want.
5
Answer returns an LLM-written response with citations at 5 dollars per thousand requests, and Deep Search performs multi-step research at 12 to 15 per thousand depending on the variant. Monitors run a query on a schedule and surface new matches at 15 per thousand.
6
Everything is reachable through the REST API with Python, JavaScript, and cURL paths, through webhooks for real-time callbacks on asynchronous Websets jobs, or through the dashboard for people who want to run and export a set without writing code.

Feature breakdown

28 features in 5 modules

Search and retrieval

The core index, priced per thousand requests with nothing gated.

Semantic web search: Describe what you want in natural language rather than in keywords and search Exa's own index of the live web, at 7 dollars per thousand requests covering the first ten results.
Contents retrieval: Pull full page text, highlights, or summaries from URLs you already hold at 1 dollar per thousand pages, which is the cheapest operation in the catalogue.
AI summaries: Generated page summaries at an additional 1 dollar per thousand pages, priced separately so you only pay when you want them.
Answer with citations: An LLM-written answer to a question with the sources it drew on, at 5 dollars per thousand requests, which is a research primitive rather than a prospecting one.
Deep Search: Multi-step research across lite, standard, and reasoning variants at 12 to 15 dollars per thousand requests, for questions a single query cannot resolve.
Additional results pricing: Results beyond the first ten cost 1 dollar per thousand, so a broad sweep is cheap once the query itself has been paid for.

Websets prospecting

The part of Exa that competes with this category, and it competes on a different axis.

Described-set queries: Express a target list as a description rather than a filter, such as agtech companies in the United States that raised a Series A, which no filter-based database in this category can represent.
Per-result verification: Each candidate is checked against rules you define before it enters the set, so what you receive has been evaluated rather than merely matched on a keyword.
Verification before enrichment: Enrichments run only on verified results, which means your enrichment spend lands on rows that passed your criteria rather than on the full candidate pool. This is the structural equivalent of not charging for a miss.
Custom enrichment specifications: Define the data points you want extracted for each result as text, numbers, dates, or booleans, rather than accepting a fixed schema of firmographic fields.
Contact enrichment: Email addresses at 0.02 dollars and phone numbers at 0.07, priced as line items rather than as a separate credit type or rationed allowance.
Asynchronous processing with webhooks: Websets jobs run asynchronously across many items with HTTP callbacks on completion, which suits a long research sweep rather than an interactive lookup.

Monitoring and agents

Standing queries and fixed-effort agent calls, both priced per request.

Monitors: Run a query on a schedule and surface new matches as they appear, at 15 dollars per thousand requests, which turns a one-off list build into an ongoing signal feed.
Agent with fixed-effort pricing: Agent calls run from 0.012 to 1.00 dollars per request depending on effort, with metered modes available, so a research task has a knowable ceiling rather than an open-ended token bill.
Agent Compute Units: Metered agent work is billed at 0.10 dollars per compute unit, published on the same page as everything else rather than quoted.
Import and enrich your own URLs: Bring a list you already have and run Exa's verification and enrichment over it, rather than being forced to start from an Exa search.
Dashboard export: Sets can be run and exported from the dashboard without writing code, which is the non-developer path.

Pricing mechanics

The most transparent commercial model in this category.

Pure pay-as-you-go: No subscription, no minimum spend, and no plan tiers. You pay for the operations you run, which is the opposite of every credit-plan vendor in this batch.
No expiring credits: Because there is no monthly allowance, there is nothing to expire. This sidesteps CUFinder's monthly reset and even Tomba's twelve-month validity by removing the concept.
20 dollars in free credits on signup: Roughly 2,800 searches, which is a substantial real evaluation rather than a token allowance.
10 dollars in free credits monthly: A recurring monthly free allowance on top of the signup grant, which effectively makes light usage permanently free.
Every operation priced separately: Search, contents, summaries, answers, deep search, monitors, agents, and each enrichment type carry their own published rate, so you can model a workload precisely before running it.
No annual commitment: The published rates require no contract at all; enterprise arrangements exist for volume discounts, custom indexes, higher rate limits, SLAs, and zero data retention, but nothing is withheld from self-serve buyers.

Developer surface

Built for code first, with the interface as a convenience.

REST API with Python and JavaScript paths: Documented API access with language examples and cURL, which is the primary way the product is consumed.
Webhooks: Real-time HTTP callbacks for asynchronous Websets jobs rather than requiring polling.
Dashboard: A web interface for running, inspecting, and exporting sets without writing code.
Enterprise options: Volume discounts, custom indexes, higher rate limits, SLAs, and zero data retention available through sales, none of which are prerequisites for using the product.
Rate limits not published per endpoint: Higher rate limits are named as an enterprise benefit, which implies default limits exist, but Exa does not publish them per endpoint on its pricing page.

Use cases

4 documented

Founder whose target market cannot be expressed as a filter

The ideal customer is a company that recently migrated off a specific competitor, which no firmographic filter in any contact database can represent.

A Websets query describes the condition in words, Exa searches and verifies each candidate against it, and only the survivors get enriched with an email at 0.02 dollars, so the research and the contact data are one operation.

Developer building a research feature into a product

The product needs to look up live information about a company at runtime, and a static contact database is the wrong shape entirely.

Search at 7 dollars per thousand and contents at 1 dollar per thousand cover it with no subscription, and the 10 dollars of free monthly credit covers development and early usage indefinitely.

Operator watching for a trigger event

Outreach converts when a target company announces a specific kind of change, and checking manually across two hundred accounts is not sustainable.

Monitors run the query on a schedule at 15 dollars per thousand requests and surface new matches as they appear, turning a one-off list into a standing signal feed without a separate intent data subscription.

Analyst enriching a list they already own

A list of five hundred company URLs exists and needs verified attributes attached, but no vendor's fixed schema matches the fields that matter.

Import the URLs, define custom enrichment specifications as text, numbers, dates, or booleans, and pay only for the operations run, with contact details added at published per-item rates.

Pricing

from $0 ($20 in free credits on signup plus $10 monthly), then $7 per 1,000 searches

Pure pay-as-you-go usage pricing with every operation individually published, no subscription, no minimum spend, no tiers, and no expiring credits.

Plan	Price	Includes
Free credits	$0 $20 on signup plus $10 every month	Roughly 2,800 searches from the signup grant Recurring $10 monthly credit Full API access No card required to begin The most substantial free allowance in this batch by a wide margin.
Search	$7 per 1,000 requests	Semantic search over Exa's own index First 10 results included Additional results at $1 per 1,000 No subscription required
Contents and Answer	$1 to $5 per 1,000 requests	Page contents at $1 per 1,000 pages AI summaries at $1 per 1,000 pages Answer with citations at $5 per 1,000 requests Priced independently of search
Deep Search and Monitors	$12 to $15 per 1,000 requests	Deep Search at $12 to $15 across lite, standard, and reasoning variants Monitors at $15 per 1,000 scheduled requests Multi-step research and standing queries
Agent	$0.012 to $1.00 per request	Fixed-effort pricing with a knowable ceiling Metered modes at $0.10 per Agent Compute Unit For research tasks beyond a single query
Enterprise	Custom quoted	Volume discounts Custom indexes Higher rate limits and SLAs Zero data retention An addition to the published rates rather than a gate in front of them; nothing is withheld from self-serve buyers.

Add-ons

Contact enrichment: email ($0.02 per email): Charged per enriched item on verified Websets results rather than as a credit type.
Contact enrichment: phone ($0.07 per phone number): Roughly three and a half times the email rate, which is the usual premium for phone data.

Billing notes

There is no subscription, no minimum spend, and no tier ladder. Every operation carries a published per-unit rate and you pay for what you run, which is the most transparent commercial model in this category.
Because there is no monthly allowance, there are no expiring credits. This sidesteps CUFinder's monthly reset entirely and is even simpler than Tomba's twelve-month validity.
New accounts receive 20 dollars in credits, roughly 2,800 searches, plus 10 dollars every month thereafter. Genuinely light usage is therefore permanently free, which no other vendor in this batch matches.
In Websets, verification runs before enrichment, so contact enrichment charges land only on results that already passed your criteria. That is structurally equivalent to the no-charge-on-a-miss rule that Hunter, Anymail Finder, Skrapp, GetProspect, SalesQL, Tomba, CUFinder, Voila Norbert, BetterContact, FullEnrich, and Dropcontact all commit to, and it puts Exa clearly apart from RocketReach, ContactOut, Datagma, Icypeas, and Enrich, which decline to state their rule. Exa does not, however, publish what happens when an enrichment itself returns nothing, which is a narrower but real gap.
Phone numbers are priced as a separate line item at 0.07 dollars rather than as a rationed credit type, so there is no allowance to run out of and no multiplier to reason about.
No annual commitment is required for any published rate. Enterprise arrangements add volume discounts, custom indexes, higher rate limits, SLAs, and zero data retention, but they are an extension of the self-serve product rather than a gate in front of it.
Default per-endpoint rate limits are not published, though higher rate limits are named as an enterprise benefit, which implies defaults exist. Confirm them before building a high-concurrency workload.

Value assessment: Compared as a contact data vendor, Exa is expensive: 1,000 emails costs 20 dollars against Tomba's 8.90, and 10,000 costs 200 against Tomba's 89. Phone numbers at 0.07 are 70 dollars per thousand and 700 per ten thousand, more than CUFinder's 299 for 10,000 mixed credits and well above SalesQL's 119 for 12,000. On that axis Exa loses to most of this batch. But that comparison misreads the product. What you are actually buying is the ability to find companies that no filter can describe, verify each one against your own criteria, and only then pay to enrich the survivors, with a research and monitoring layer on the same bill and no subscription underneath any of it. Nothing else in this category does that at any price. For a team whose bottleneck is knowing who to contact rather than how to reach them, 7 dollars per thousand searches with 10 dollars free every month is exceptional value. For a team that already has the list, it is simply the wrong tool.

Strengths & limitations

Strengths

The only product in this category that answers a described query rather than a filtered one, which makes previously unbuildable target lists buildable.
Verification runs before enrichment in Websets, so contact spend lands only on results that already passed your criteria.
Pure pay-as-you-go with no subscription, no minimum spend, no tiers, and nothing to expire, which is the cleanest commercial model in this batch.
Every operation individually priced and published, from 1 dollar per thousand page fetches to 0.07 per phone number, so any workload can be modelled precisely before it is run.
20 dollars in free credits on signup plus 10 dollars every month makes light usage permanently free, a far more generous allowance than any competitor here.
Monitors turn a one-off list build into a standing signal feed without a separate intent data subscription.
Extremely well capitalised after a 250 million dollar Series C at a 2.2 billion valuation, so the API is not going to disappear underneath an integration.

Limitations

Contact data is expensive per unit: 0.02 dollars an email is more than twice Tomba's rate, and 0.07 a phone is well above CUFinder's or SalesQL's effective rates.
It is not a contact database, so there is no fixed people index to filter by title and seniority and no guarantee that a given person is findable at all.
No match rate is published for either email or phone enrichment, and Exa does not state what happens when an enrichment itself returns nothing.
Non-technical users will struggle; the dashboard exists but the product is designed to be called from code and the documentation is written for developers.
No CRM, no sequencer, no lead lists, and no workflow layer, so everything downstream of the data is your responsibility.
Default per-endpoint rate limits are not published, which matters for any high-concurrency workload.
Compliance framing is that of a search company rather than a contact data vendor, so there is little published about where enrichment contact data originates, what legal basis supports it, or how an individual objects.
Websets jobs are asynchronous and can be slow relative to a database lookup, because the product is doing research rather than reading a row.

Head-to-head comparisons

5 alternatives

Exa vs CUFinder

from $0 (Free, 15 credits), then $49 per month (Lite, 1,000 credits)

Opposite ends of the same problem. CUFinder sells a pre-built database, buying signals, and a finished interface on a monthly subscription with expiring credits, at 0.049 to 0.0299 per successful match of any type. Exa sells live semantic search and per-item enrichment with no subscription and nothing to expire, at 0.02 per email and 0.07 per phone on top of 7 dollars per thousand searches. Take CUFinder if a salesperson opens the tool each morning; take Exa if code does the work and the list cannot be expressed as a filter.

Full Exa vs CUFinder comparison

Exa vs Tomba

from $0 (25 free searches per month, plus 75 free API credits on signup), then $8.90 per 1,000 credits

Both are API-first with flat published usage pricing and no tier ladder, and both are honest about what they charge for. Tomba is far cheaper for the specific job of turning a name and a domain into a verified email, at 0.0089 against Exa's 0.02, and ships ten SDKs across fourteen endpoints. Exa cannot be beaten on discovering which companies to target in the first place. The sensible answer for many teams is both: Exa to build the set, Tomba to resolve the addresses.

Full Exa vs Tomba comparison

Exa vs Clay

from Free plan; paid from $149/mo

Clay is the orchestration layer that most teams reach for when a list needs research plus waterfall enrichment, with bring-your-own-key support across dozens of providers and a spreadsheet interface non-technical operators can use. Exa is one of the primitives such a stack is built on, cheaper per operation and far more flexible, but it requires you to build the workflow yourself. Take Clay if you want the workflow; take Exa if you want the capability and intend to wire it up.

Full Exa vs Clay comparison

Exa vs Apollo.io

from Free plan; paid from $49/user/mo

Apollo is a complete sales platform with a large searchable database, a sequencer, and a usable free tier, and for a conventional filter-based list build it will be faster and cheaper than Exa every time. Exa is not a competitor to it so much as a complement, handling the target definitions Apollo's filters cannot express. Most small teams should own Apollo or an equivalent and reach for Exa when a list proves impossible to describe with dropdowns.

Full Exa vs Apollo.io comparison

Exa vs TheirStack

from Free tier on signup, then $49 per month for 1,500 credits or $109 once for 1,000 credits

TheirStack answers a narrow research question, which companies use which technologies and are hiring for what, from a maintained dataset with published pricing. Exa answers arbitrary research questions from the live web at a lower unit price but with more work required to shape the query and validate the output. Take TheirStack when your question is exactly the one it was built for; take Exa when it is not.

Full Exa vs TheirStack comparison

Implementation & onboarding

Setup time: Twenty minutes to a first useful result: create an account, take the 20 dollars of free credit, and call search from Python or JavaScript. A production Websets pipeline with verification rules and custom enrichments is a few days of work rather than an afternoon.
Learning curve: Moderate to high by this category's standards, and the difficulty is conceptual rather than technical. Writing a good described query and a good verification rule is a skill, and a badly specified set returns plausible-looking results that quietly fail your criteria. Teams used to filter dropdowns will need to change how they think about list building.
Onboarding: Entirely self-serve with no sales contact at any published rate. Enterprise arrangements exist for volume discounts, custom indexes, higher rate limits, SLAs, and zero data retention, but nothing is withheld from a self-serve account.
Migration notes: There is nothing to migrate in, because Exa holds no data of yours beyond what you generate. On the way out, sets export from the dashboard and everything is retrievable through the API. The genuine lock-in risk is not the data but the query logic: a well-tuned Websets specification represents real work that does not transfer to any other vendor, because no other vendor accepts a query of that shape.

Platform, API & security

Platforms: REST APIPython and JavaScript clientsWeb dashboardWebhooks
API: A REST API covering search, contents, answer, deep search, websets, monitors, and agents, with Python, JavaScript, and cURL paths and webhook callbacks for asynchronous jobs. Default per-endpoint rate limits are not published, though higher rate limits are offered as an enterprise benefit, which implies defaults exist and should be confirmed before building a high-concurrency workload.
Compliance: Zero Data Retention available through enterprise arrangements
Data residency: No customer-selectable regional hosting is advertised on the self-serve product; custom indexes are an enterprise option.
SSO: Not advertised on the self-serve product.
Security notes: Exa's compliance posture is that of a search company: it indexes the public web and offers zero data retention as an enterprise option. It publishes considerably less than a dedicated contact data vendor about where enrichment contact details originate, what legal basis supports processing personal data, or how an individual objects to inclusion. Buyers using Exa's contact enrichment for European outreach should request that documentation specifically, since the general search framing does not cover it.

Support & resources

Channels: Email supportSales contact for enterprise arrangements
Documentation: Extensive developer documentation at exa.ai/docs covering search, contents, answer, deep search, websets, enrichments, monitors, agents, and webhooks, with worked examples in Python, JavaScript, and cURL.
Community: Developer-oriented presence through documentation, the company blog, and the AI engineering community rather than a formal user forum.

Company

Founded: 2021
Headquarters: San Francisco, California, United States
Ownership: Venture-backed, operating as Exa Labs Inc.
Founders: Will Bryk, Jeff Wang
Employees: Not disclosed
Funding: Approximately $335M raised, including an $85M Series B led by Benchmark in September 2025 at a $700M valuation and a $250M Series C led by Andreessen Horowitz in May 2026 at a $2.2B valuation.

Funding history

Round	Amount	Year	Notes
Series B	$85M	2025	Led by Benchmark with Lightspeed, Y Combinator, and NVentures participating, at a $700M valuation.
Series C	$250M	2026	Led by Andreessen Horowitz at a $2.2B valuation, roughly tripling the company's value in eight months.

Timeline

2021Founded in San Francisco by Will Bryk and Jeff Wang on the premise that search built for humans is the wrong shape for software.
2023Ships a semantic search API over its own web index, positioned for AI systems rather than human browsing.
2024Adds contents retrieval, answers with citations, and the first agentic research capabilities on top of the base index.
2025Raises an $85M Series B led by Benchmark at a $700M valuation and launches Websets, a high-compute product that searches, verifies, and enriches described sets.
2026Raises a $250M Series C led by Andreessen Horowitz at a $2.2B valuation, publishing per-operation pricing across search, deep search, monitors, agents, and contact enrichment.

Integrations

REST API with Python and JavaScript clients
Webhooks for asynchronous Websets jobs
Dashboard export
Import your own URLs for verification and enrichment
Custom indexes on enterprise arrangements

Frequently asked questions

10 questions

What is Exa and is it a B2B contact database?

Exa is a web search and data API built for AI systems, not a contact database. It sells semantic search over its own index of the live web, page content retrieval, agentic research, and Websets, which runs a described prospecting query, verifies each result against criteria you set, and enriches the survivors with contact details. The contact data is the end of a research pipeline rather than the product, which makes Exa a fundamentally different purchase from Apollo, Hunter, or Skrapp.

How much does Exa cost?

Everything is pay-as-you-go with no subscription and no minimum spend. Search is 7 dollars per thousand requests, page contents 1 dollar per thousand, AI summaries another 1, answers with citations 5, deep search 12 to 15, and monitors 15. Agent calls run 0.012 to 1.00 dollars each. Contact enrichment is 0.02 dollars for an email and 0.07 for a phone number. New accounts get 20 dollars in credits plus 10 dollars every month thereafter.

Does a failed lookup cost anything?

In Websets, verification runs before enrichment, so your contact enrichment spend lands only on results that already passed your criteria rather than on the whole candidate pool. That is structurally equivalent to the charge-on-success rule that Hunter, Anymail Finder, Skrapp, GetProspect, SalesQL, Tomba, CUFinder, Voila Norbert, BetterContact, FullEnrich, and Dropcontact all commit to, and it separates Exa from RocketReach, ContactOut, Datagma, Icypeas, and Enrich, which will not state their rule. Exa does not publish what happens when an enrichment itself returns nothing, so that narrower question is worth asking directly.

Are phone numbers priced separately from emails?

Yes, as separate line items rather than as a rationed credit type. Email enrichment is 0.02 dollars per item and phone enrichment is 0.07, roughly three and a half times the rate. Because there is no credit pool and no monthly allowance, there is nothing to run out of and no multiplier to reason about, which is simpler than Kaspr's three separate meters or Tomba's five-credits-per-phone rule.

What does Exa cost per verified email and per mobile number at 1,000 and 10,000 lookups?

Emails are 20 dollars for 1,000 and 200 for 10,000. Mobile numbers are 70 dollars for 1,000 and 700 for 10,000. Both figures sit above the field: Tomba charges 8.90 and 89 for the same email volumes, and CUFinder charges 299 for 10,000 mixed credits. Those comparisons are fair on unit price and misleading on purpose, because Exa's cost also buys the research that produced the list, which the cheaper vendors cannot do at any price.

Does Exa publish a match rate for contact enrichment?

No, and this is the significant gap for anyone treating Exa as a contact data vendor. It makes no public commitment about how often an email or phone enrichment succeeds on a given segment, and it publishes no accuracy figure of the kind CUFinder, Skrapp, and Voila Norbert advertise, however loosely those are backed. The 20 dollars of free signup credit plus 10 dollars a month is a substantial real budget for testing this yourself before committing.

Where does Exa's contact data come from and is it usable for European outreach?

Exa's published compliance framing is that of a search company indexing the public web, with zero data retention offered as an enterprise option. It says much less than a dedicated contact vendor about where enrichment contact details originate, what GDPR legal basis supports processing them, or how an individual objects to inclusion. If you intend to use Exa's contact enrichment for outreach to people in the EU, request that documentation specifically, because the general search-company framing does not cover it and your own legitimate interest assessment remains your responsibility.

Do Exa credits expire, and is an annual commitment required?

Neither. There is no subscription, no monthly allowance, and therefore nothing to expire, which sidesteps CUFinder's monthly reset entirely and is even simpler than Tomba's twelve-month validity. No annual commitment is required for any published rate. Enterprise arrangements add volume discounts, custom indexes, higher rate limits, SLAs, and zero data retention, but they extend the self-serve product rather than gating it.

What are Exa's API rate limits?

Default per-endpoint rate limits are not published, which is the notable omission in otherwise exemplary pricing transparency. Higher rate limits are listed as an enterprise benefit, which confirms that defaults exist. Any team planning a high-concurrency workload should establish them before building, either by asking or by testing, in the same way they would with Tomba or CUFinder.

Should a small sales team buy Exa instead of a normal prospecting tool?

Usually not instead, sometimes as well. If your bottleneck is knowing how to reach people you can already name, buy Tomba, Skrapp, or GetProspect and pay a fraction of Exa's per-email rate. If your bottleneck is that your ideal customer cannot be described with dropdown filters, Exa is the only tool in this category that can express the question, and the monthly free credit means testing that costs nothing. Many teams end up using Exa to build the set and a cheaper finder to resolve the addresses.

Editorial verdict

Exa is the odd one out in this category and the most interesting thing in it. It does not sell a contact database, it sells the ability to ask the live web a question no filter can express, verify each answer against your own criteria, and attach an email or a phone number to the results that survive. Priced purely per operation with 10 dollars free every month, no subscription, and nothing that expires, it is the most transparent commercial model here by a distance, and a Series C at a 2.2 billion valuation means the API will still be there in three years. The honest caveats are that contact data costs more than twice Tomba's rate, there is no published match rate, the compliance documentation is written for a search company rather than a data vendor, and a non-technical salesperson will not enjoy using it. Buy Exa when the problem is deciding who to contact. Buy something cheaper when the problem is only how to reach them, and consider running both, which is what the teams getting the most out of it actually do.

Written by the SaaSTracker editorial team. Awards, when shown, are judged against the published criteria in our methodology.

Awards & badges

1 held

Exa holds 1 award from the SaaSTracker editorial program. Badges may be displayed by the vendor; each embed links back to this profile.

Innovation · B2B Data Providers

“Semantic search over the live web with contact data attached made prospecting a language problem rather than a database subscription.”

Embed code

<a href="https://saastracker.org/products/exa-ai"><img src="https://saastracker.org/badges/embed/exa-ai/innovation-summer-2026.svg" width="180" height="180" alt="Exa: Innovation, B2B Data Providers. SaaSTracker Summer 2026 Awards." /></a>

Download: SVG · PNG