Vulnerability scan vs penetration test: what each one hands you
Someone has told you that you need a pentest, and you have found scans that finish in minutes sold next to tests that run for days. They are different purchases: a vulnerability scan gives you a list of things that might be exploitable, and a penetration test gives you findings that were exploited, with the evidence attached. A web app with real users needs both, and if you were asked for a pentest, buy a pentest, because a scan report won’t answer the request.
If you only want quick, regular coverage of known issues, a scan is enough. Otherwise, run scans on a schedule and book a pentest before launch and after big changes.
Vulnerability scan vs penetration test at a glance
The line between a vulnerability scan and a penetration test is exploitation. NIST’s testing guide, SP 800-115 (September 2008, still its final version as of November 2026), draws it in one sentence: “While vulnerability scanners check only for the possible existence of a vulnerability, the attack phase of a penetration test exploits the vulnerability to confirm its existence.”
Here is how the two compare, with our own pentest as the third row:
| Option | How it works | What you get | Time | Limit |
|---|---|---|---|---|
| Vulnerability scan | Automated tools match what they find against known vulnerabilities | A ranked list of potential issues | Seconds to minutes per host | Flags possible issues only; false positives |
| Penetration test | A manual process, helped by tools, that tries to exploit weaknesses | Each issue verified, with how and how far it can be exploited | Days or weeks | Labor-intensive, high cost, needs expert testers |
| Antlion AI pentest | Our own AI models attack one confirmed domain from a throwaway sandbox | A written report where each finding was proven exploitable, plus one re-test | 3 to 4 hours, report within 4 to 24 hours | One domain, no subdomains; no DDoS-style attacks |
The first two rows follow NIST SP 800-115 and section 2.1 of the Penetration Testing Guidance from the PCI Security Standards Council, or PCI SSC (version 1.1, September 2017). The third comes from our own pricing and process.
What each one does and what it hands you
A vulnerability scan ends in a ranked list of possible issues, a penetration test in verified findings, and Antlion’s AI pentest in a report where each finding was proven exploitable.
Vulnerability scan
In NIST’s description, a vulnerability scan identifies the operating systems and major applications on each host, then matches them with the known vulnerabilities in the scanner’s database. For web apps, OWASP’s page on vulnerability scanning tools describes scanners as automated tools that probe an application from the outside for flaws such as cross-site scripting and SQL injection.
What you get back is a list. PCI SSC’s Penetration Testing Guidance says a scan reports potential risks from known vulnerabilities, ranked by CVSS base score. For example, a scan that finds an environment file on your web server might report it like this (an illustration, not real scan output):
MEDIUM https://example.com/.env Possible sensitive file exposed
The line is accurate, and it doesn’t tell you what is in the file or what an attacker can do with it. NIST is blunt about the limits: scanners “can have a high false positive error rate”, and the risk level a scanner assigns may not reflect your actual risk.
Penetration test
NIST defines a penetration test as security testing in which assessors “mimic real-world attacks to identify methods for circumventing the security features of an application, system, or network”. The attack is aimed at your own system and run with your approval, to find out what an intruder could do.
NIST’s four phases are planning, discovery, attack and reporting. A scan lacks the attack phase, where testers chain weaknesses to gain more access than any single one gives.
So what does a pentest finding look like? Here is one from our sample report, shortened. The target and the finding are fictional and show the format:
ANT-002 Environment file readable from the browser (High)
Affected asset: https://example.com/.env
Evidence
GET /.env HTTP/1.1
Host: example.com
HTTP/1.1 200 OK
Content-Type: text/plain
Content-Length: 412
DATABASE_URL=postgres://app:[redacted]@db:5432/app
SESSION_SECRET=[redacted]
PAYMENT_API_KEY=[redacted]
Impact
Anyone who downloads the file obtains the database credentials and
the application secrets. Combined with ANT-003, the database can be
read and changed directly.
Remediation
Treat every secret in the file as compromised and rotate it.
(second of three steps)
How to confirm the fix
A request to https://example.com/.env should return 403 or 404.
The file was downloaded, so the finding shows the request, the secrets that came back, and how it combines with ANT-003, a database that accepts remote connections. PCI SSC’s Penetration Testing Guidance (section 5.3.1) counts screenshots, raw tool output and data obtained through exploitation as evidence, and says that merely reporting lists of vulnerabilities “does not meet the intent of the penetration test”.
A pentest has limits too. NIST calls it labor-intensive, and the OWASP Web Security Testing Guide, citing Gary McGraw, warns that it can only identify a small sample of a system’s risks.
Antlion’s AI pentest and its limits
Antlion is an AI pentest for one website. You confirm the domain is yours, and our own pretrained models attack it.
Antlion’s agents work in parallel and test your site for more than 25,000 recently disclosed vulnerabilities. Its other tools test the endpoints a server leaves open: admin panels such as Dokploy, databases, APIs and files that should not be public.
An Antlion test counts as a pentest because each finding in the report was proven exploitable. The test takes 3 to 4 hours, the written report arrives within 4 to 24 hours, and one re-test is included.
Can software run a pentest at all? PCI SSC’s Penetration Testing Guidance, published in 2017, calls a penetration test a manual process that may use automated tools. If an assessor has to accept your report, ask them before you buy whether an AI-run test qualifies.
Antlion has two more limits. One pentest covers one confirmed domain, so subdomains are not tested, and we run no DDoS-style attacks, so it won’t tell you how your site handles a flood of traffic.
As of November 2026 the price is $299 (regular $499).
What to ask a provider before you pay
The name on the invoice won’t tell you whether you are buying a scan or a pentest, so ask. The first five questions are adapted from the report checklist (section 5.4) and the evidence section of PCI SSC’s Penetration Testing Guidance, and the sixth is ours:
- Do you try to exploit each finding, or do you report what a tool flags?
- What evidence comes with each finding?
- Which parts of the test are automated, and which are done by hand?
- What exactly is in scope, and what is left out?
- Is a re-test included once we have fixed things?
- Can I see a sample report before I pay?
The first two settle it. If the answer is “we run the tool and send you the results”, you are buying a scan, which is fine at a scan price.
Then read one finding in their sample report. A pentest finding shows what the tester did and what came back; a scan finding shows a title and a severity.
When a scan is enough and when you need a pentest
A scan is enough for routine coverage, and a pentest is needed when someone has to know what an attacker can do. NIST pairs them: regularly scheduled vulnerability scanning, “interspersed with periodic penetration testing”, can help prevent many types of attacks.
A scan is enough when:
- You want outdated software, missing patches and misconfigurations caught between releases, which is what NIST says scanning identifies.
- The requirement in front of you names a scan. PCI DSS v3.2, for instance, expects scans at least quarterly and after significant changes, according to PCI SSC’s Penetration Testing Guidance.
You need a pentest when:
- A customer or a standard asks for a penetration test. The same PCI SSC guidance expects one at least annually and upon significant changes, and tells you to check the current version of the standard.
- You are about to launch, or you just shipped a new login flow, payments or an admin panel.
- A scan returned a long list and nobody on your team can say which items are real.
- The risk is in a combination. Say your app leaks user IDs in one place and skips an ownership check in another: together they expose every account.
If you have tested nothing yet, run a scan this week, then book a pentest before your next launch. A vulnerability scan tells you where to look, and a penetration test shows what an attacker can do today, with proof.
If your app lives on one domain and you want every finding proven before you fix it, that is the test we offer. Request a pentest