Penetration Testing Cost
Pen Test Pricing by Type, AI and LLM Penetration Testing, and How to Buy

A single pen test usually costs $5,000 to $50,000, and a red team engagement $40,000 to $200,000. Send us your scope once: we take it to the testers that fit and negotiate the price.

Price My Pen Test
For security, IT and compliance leaders Paid by the provider, so it costs you nothing Not tied to any one testing firm

Free Pen Test Scoping Review: send it before you request quotes. You'll hear back within 24 hours.

Penetration testing cost for a single test usually runs $5,000 to $50,000, set mostly by the type of test and how much is in scope. eSecurity Planet's average of 10 published sources puts network tests at $9,900 to $53,700 and application tests at $8,900 to $34,600, while red team engagements run $40,000 to $200,000 or more. Testers bill about $157 an hour at the median of more than 400 penetration tester rates on federal GSA schedules (MCA calculation from GSA CALC). This page prices each test type with sources, shows what an annual testing program costs a 500-person and a 1,000-person company, covers AI and LLM penetration testing, and gives the scoping questions buyers put in an RFP. We source and negotiate penetration testing and the managed security services around it for mid-size and large companies, and the provider you choose pays us. See how we work.

FEDERAL TESTER RATE
$157
median per hour across 400+ GSA penetration tester rates (MCA calculation, GSA CALC)
NETWORK TEST
$9,900 to $53,700
average low to high across 10 sources (eSecurity Planet)
TESTING PLATFORMS
$30,000 to $41,000
median yearly spend across four testing platforms (Vendr)
US BREACH AVERAGE
$11.5M
IBM 2026, with a 247-day lifecycle (Baker Donelson summary)
Penetration testing cost by type, per test
Typical price range for one test of moderate scope, in US dollars. Logarithmic scale, so small and large engagements both read.
Social engineeringPhishing and vishing: $4,000 to $10,000
External network$5,000 to $20,000
API$5,000 to $20,000
Web application$5,000 to $30,000
Mobile application$5,000 to $30,000
Internal network$7,000 to $35,000
Cloud (AWS, Azure, GCP)$10,000 to $40,000
AI and LLM application$6,000 to $50,000
Red team$40,000 to $200,000
$2K$5K$10K$25K$50K$100K$250K
Scoped penetration testsAI and LLM application testRed team engagement

Sources: planning ranges published by US testing firms Blaze Information Security, Compass IT Compliance and CyCognito (2025 to 2026), and SecurityWall for AI and LLM tests; cross-checked against eSecurity Planet's 10-source averages and Vendr buyer data, which sets the red team upper bound. Each range is the span the published sources share for a moderate scope; very large estates price above it. Figures as of October 2026. Free to cite with a link to this page.

Free Pen Test Scoping Review
Send what needs testing: IP counts, applications and APIs, cloud accounts, the compliance driver and when the report is due. You'll hear back within 24 hours. From there, we work out what the test should cost at your scope and which testers fit. Send it before you request quotes: once a testing firm has priced you directly, it will not register an advisor, and you negotiate alone.
Price My Pen Test

Penetration Testing Cost by Type of Test

Each test type is priced by its own unit: IP addresses for networks, roles and pages for applications, endpoints for APIs, accounts for cloud, people for social engineering. The table gives the typical range, what drives it, and how long the testing usually takes.

Test typeTypical cost per testPriced byTesting time
External networkinternet-facing hosts, VPN, mail, firewalls$5,000 to $20,000; up to about $30,000 past 100 IPsLive IP addresses and exposed servicesAbout one week
Internal networkassumed breach inside the office or data center$7,000 to $35,000Sites, network segments, Active Directory sizeAbout one week
Web application$5,000 to $30,000 per applicationUser roles, pages and functions, authenticationOne to three weeks
APIREST, GraphQL, partner integrations$5,000 to $20,000Endpoints and roles; documentation suppliedOne to two weeks
CloudAWS, Azure, Google Cloud configuration and attack paths$10,000 to $40,000Accounts or subscriptions, services in use, identity setupOne to three weeks
Mobile applicationiOS, Android and their back end$5,000 to $30,000Platforms, app complexity, back-end APIsOne to two weeks
Social engineeringphishing, phone pretexting, physical$4,000 to $10,000 for phishing and phone campaignsNumber of people, campaigns, sitesSet by the campaigns
AI and LLM applicationchatbots, RAG, agents with tools$6,000 to $50,000Model integrations, tools and data the model can reachOne to four weeks
Red teamgoal-based, stealthy, full organization$40,000 to $200,000Objectives, duration, techniques allowedTwo to eight weeks or more

Ranges and durations as in the chart sources above. CISA's no-cost Risk and Vulnerability Assessment runs one week of external and one week of internal testing, a useful benchmark for network test length.

What penetration testing costs per hour and per day

Most quotes are built from tester-days, even when the price is fixed. Two government price sources show the rate underneath:

US federal schedules

$126 to $195

an hour for the middle half of more than 400 penetration tester labor rates on GSA schedules, median about $157, full range about $57 to $321. Most sit on the HACS cybersecurity SIN, where penetration testing is its own subgroup. MCA calculation from GSA CALC ceiling rates.

UK government G-Cloud

£900 to £1,500

a tester-day in supplier price documents: £1,200 for most test types and £1,500 for hardware at one supplier; £900 to £1,100 at another. Excluding VAT.

What a tester-day costs

$1,253

at the GSA median: $156.62 x 8 hours (MCA calculation). A 10-day test is about $12,530 and a 20-day test about $25,060 before travel, reporting extras or retests, which is why most tests land inside the ranges above.

WHAT THAT MEANS
Ask every tester for the number of tester-days behind a fixed price. Two quotes for "a web application test" can differ by 3x simply because one assumes 5 days and the other 15. Once days are on the page, quotes compare line for line, and the rate per day becomes something you can negotiate.

Penetration Testing Pricing Models

Fixed price per test

The most common model: one price for a defined scope and a report. Vendr's data puts project-based tests at $15,000 to $75,000 (Vendr). Fixed bids carry contingency, so a tight scope gets a better price.

Time and materials

Billed by the tester-day or hour at the rates above. Suits open-ended work such as a red team or a complex internal estate. Put a cap on days in writing.

Pen testing as a service

A yearly subscription or credit bank for repeat testing. Vendr estimates credits at $800 to $1,500 each, with a web application test using 10 to 20 (Vendr), and annual programs at $50,000 to $200,000 or more.

Automated testing

Software that runs network attack paths on a schedule, priced by IP or host: £12 per external IP and £8 per internal IP on G-Cloud 15; $4,000 to $10,000 a year for 1 to 150 external hosts on AWS Marketplace. It adds frequency; it does not replace a tester for applications.

What buyers pay testing platforms

Testing platformMedian per yearRange in buyer dataPurchases
Cobaltcredit-based pen testing as a service$30,000$8,937 to $87,840194 (Vendr)
NetSPIproject tests and pen testing as a service$37,688$13,200 to $156,44347 (Vendr)
Bugcrowdmanaged pen tests and bug bounty$39,196$10,000 to $80,563192 (Vendr)
HackerOnepen tests and bug bounty$41,000$19,904 to $129,802319 (Vendr)

Median deal size reflects each buyer set's size and scope, not a like-for-like price. Platform totals often include bug bounty and attack surface services alongside pen testing.

What Drives Pen Test Pricing

IP addresses and sites

Network tests price by live hosts and services, not total address space. Count what responds, and list each site and segment that internal testing must reach.

Applications and roles

Each application is priced separately, and each user role adds testing: an admin, a customer and a partner role is three sets of permissions to try to break.

Black, gray or white box

Giving testers credentials, documentation or source code costs more per day but finds more per day. Gray box, with test accounts, is the usual middle ground.

Retest

A retest confirms your fixes. Some suppliers include it: one G-Cloud price document retests high and critical findings free up to one day's effort (G-Cloud 15). Get it in the quote.

Timing

Out-of-hours testing costs more: 1.5 times the day rate on weekday evenings and 2 times on weekends in the same price document. Rush dates before an audit narrow your choice of testers.

Compliance reporting

A report written for a PCI assessor, a federal authorization or a CMMC assessor needs specific evidence and attestation, which adds days to the engagement.

Which Rules Require Annual Penetration Testing

Most companies buy a pen test because a regulator, an auditor, a customer or an insurer asked for one. The requirement sets the frequency and often the scope, so it belongs at the top of the scoping document.

DriverWhat it requiresFrequencySource
PCI DSS v4.0.1Requirement 11.4Internal and external penetration testing, by a qualified internal or external tester with organizational independence; segmentation testing where segmentation limits scopeAt least every 12 months and after significant change; segmentation every 6 months for service providersPCI SSC
NYDFS 23 NYCRR 500.5New York financial servicesPenetration testing from inside and outside the system boundaries by a qualified internal or external partyAt least annuallyNY DFS
CMMC Level 3CA.L3-3.12.1ePenetration testing using automated scanning tools and ad hoc tests by subject matter expertsAt least annually or after significant security changes32 CFR 170.14
FedRAMPcloud services sold to federal agenciesThird-party assessor test across six mandatory attack vectorsBefore authorization, then at least every 12 monthsFedRAMP guidance
HIPAA Security Ruleproposed updateProposed: penetration testing and vulnerability scanning for covered entities and business associates; final action now projected for July 2027 (McDermott)Proposed: test every 12 months, scan every 6 monthsFederal Register
SOC 2No explicit requirement in the criteria, but a test is widely used as evidence for monitoring criteria such as CC4.1 and customers ask for itUsually annual, by customer or auditor expectationCloud Security Alliance
Cyber insurance and customersSet by each insurer's application or each customer's security reviewAs asked; keep the latest report and retest letter readyYour policy and contracts

Testing in a regulated environment usually travels with the hosting decision. Cardholder, health and defense data often sit in compliance colocation, a GCC High tenant or a CMMC enclave, and the pen test scope should follow those boundaries. Healthcare teams can see how the same rules shape HIPAA colocation, and NYDFS-covered firms how they shape financial services colocation.

What an Annual Testing Program Costs at 500 and 1,000 Employees

The table builds a typical yearly program from the per-test ranges above. The 500-employee company has one main site, two customer-facing web applications, one public API and one cloud environment. The 1,000-employee company has two sites, four web applications, two APIs, one cloud environment, a mobile app and one AI assistant in production.

TestPer test500 employees1,000 employees
External network$5,000 to $20,0001 test: $5,000 to $20,0001 test: $5,000 to $20,000
Internal network$7,000 to $35,0001 site: $7,000 to $35,0002 sites: $14,000 to $70,000
Web application$5,000 to $30,0002 apps: $10,000 to $60,0004 apps: $20,000 to $120,000
API$5,000 to $20,0001 API: $5,000 to $20,0002 APIs: $10,000 to $40,000
Cloud$10,000 to $40,0001 test: $10,000 to $40,0001 test: $10,000 to $40,000
Mobile application$5,000 to $30,000None1 app: $5,000 to $30,000
Social engineering$4,000 to $10,0001 campaign: $4,000 to $10,0001 campaign: $4,000 to $10,000
AI and LLM application$6,000 to $50,000None1 app: $6,000 to $50,000
Annual programmidpoint in brackets $41,000 to $185,000 ($113,000)$74,000 to $380,000 ($227,000)

MCA calculation: number of tests x the per-test range, summed. 500 employees: $5,000 + $7,000 + $10,000 + $5,000 + $10,000 + $4,000 = $41,000 low; $20,000 + $35,000 + $60,000 + $20,000 + $40,000 + $10,000 = $185,000 high. 1,000 employees: $5,000 + $14,000 + $20,000 + $10,000 + $10,000 + $5,000 + $4,000 + $6,000 = $74,000 low; $20,000 + $70,000 + $120,000 + $40,000 + $40,000 + $30,000 + $10,000 + $50,000 = $380,000 high. Retests and a red team are extra.

WHAT THAT MEANS
The midpoints, about $113,000 and $227,000 a year, sit inside the $50,000 to $200,000 or more that Vendr sees for annual pen testing programs. Hiring instead costs about $188,600 a year for one in-house tester at the BLS information security analyst median (MCA calculation: $129,180 x 1.46 benefits ratio), and PCI still expects organizational independence from the people who run the systems. Adding a red team every year or two puts another $40,000 to $200,000 on top. Set against the $11.5 million US breach average, the program is a small line; the gap between the low and high end is where negotiation pays.
Budgeting a year of testing?
Send us the list of what must be tested and why, and we price the whole program across the testers that fit, not one test at a time. The provider you choose pays us, so it costs you nothing. If you have already spoken to one testing firm, tell us which; we can still run the rest of the market.
Price My Testing Program

Why Buyers Come to Us First

No cost to you

The provider you choose pays us, so the scoping help and the negotiation cost you nothing. You pay the tester's price, and our job is to make that price lower.

One scope, the testers that fit

You write the scope once. We take it to the testing firms that fit your test types and compliance driver, instead of you running separate scoping calls with each.

Price leverage

We know what each test type should cost from the ranges on this page, and we negotiate tester-days, day rate, retests and a multi-test program price, not just the headline number.

Not tied to one firm

We are not tied to any one testing firm or platform, and we recommend on fit for your scope and your auditor. If it isn't a fit, we will say so.

Timing

Once you ask a testing firm for pricing directly, it will not register an advisor on the deal, and you negotiate alone. Send the scope first and we can run every firm that fits.

AI Penetration Testing and LLM Penetration Testing

AI penetration testing, also called LLM penetration testing, attacks an application built on a large language model the way a user, a document or a connected system could. It tests the model's instructions, the data it can retrieve, the tools it can call and what it does with its output. The checklist most testers work from is the OWASP Top 10 for LLM Applications (2025 edition):

LLM01 Prompt InjectionLLM02 Sensitive Information DisclosureLLM03 Supply ChainLLM04 Data and Model PoisoningLLM05 Improper Output HandlingLLM06 Excessive AgencyLLM07 System Prompt LeakageLLM08 Vector and Embedding WeaknessesLLM09 MisinformationLLM10 Unbounded Consumption

What an AI pen test covers

Prompt injection

Direct attempts to override the system's instructions, and indirect injection hidden in web pages, emails or documents the model reads. This is the first item on the OWASP list for a reason: it is how most other failures start.

Data exposure

Whether the model leaks its system prompt, other users' data, or documents from a retrieval (RAG) store that the user should not see, and whether access controls hold at the vector database.

Agents and tools

What happens when the model can send email, query systems or run code: privilege bypass, actions taken without approval, and chains from one tool to another. This is where excessive agency shows up.

The application around it

The web front end, APIs, identity and cloud hosting still get tested the usual way. Many AI findings are ordinary application flaws reached through a new interface.

What AI and LLM penetration testing costs

Published ranges run about $6,000 for a single chatbot with no tools to $50,000 or more for an agent system with retrieval, tools and compliance reporting, with 5 to 25 tester-days of work across that span (planning ranges from testing firm SecurityWall, 2026). Government rate cards show the premium: one UK supplier's G-Cloud 15 price document lists AI penetration testing at £1,600 a day against £1,200 for its standard CREST testing, about a third more. On AWS Marketplace, a manual AI and LLM test is quoted privately with engagements of one to two weeks, an annual AI testing platform lists at $100,000 for 12 months, and automated AI red teaming tools start at $500 a month for 5 million tokens of attacks.

MCA CALCULATION
At £1,600 a day, the G-Cloud AI rate, 5 to 25 tester-days is £8,000 to £40,000, which lines up with the published $6,000 to $50,000 range. Scope by what the model can reach: a chatbot answering from public pages sits at the low end, an agent with access to customer records and the ability to act sits at the top. Teams hosting their own models on AI and GPU infrastructure add the inference stack to scope.

AI red teaming services

AI red teaming goes past a scoped test: a team pursues a goal, such as getting the assistant to move money or reveal a client file, using any path, and often tests harmful-output and bias behavior alongside security. UK G-Cloud lists AI red teaming within AI security testing services, and it is priced like a red team: by days and objectives. Buy an AI pen test first, then red team once the obvious holes are closed. Data the model can reach is the bigger exposure, which is where a DSPM review of what sits in your retrieval stores pays off.

Testing an AI assistant or agent?
Tell us what the model can read, which tools it can call and who uses it. We put the scope to the testers that fit AI work and negotiate the one you choose. If an ordinary application test covers it, we will say so.
Price My AI Pen Test

Vulnerability Scan vs Penetration Test vs Red Team

 Vulnerability scanPenetration testRed team
What it doesAutomated check for known weaknessesTesters exploit weaknesses in a defined scope to show real impactGoal-based attack on the whole organization, testing detection and response
Who knowsEveryoneIT and security teamsA small control group only
How oftenContinuous to quarterlyAnnually and after major changeEvery one to two years, once basics are strong
Typical costLow; CISA offers vulnerability scanning as a service$5,000 to $50,000 per test$40,000 to $200,000
SatisfiesScanning rules, not pen test rulesPCI 11.4, NYDFS 500.5, CMMC Level 3, FedRAMPBoard and regulator questions on detection

Red team services: cost and when you need one

Red team services run $40,000 to $200,000 or more, with Vendr placing large-firm engagements at $75,000 to $200,000 or more (Vendr), and UK G-Cloud listing red team days at £1,800 against £1,200 for standard testing at the same supplier. A red team answers a different question from a pen test: not "what is broken" but "would we notice, and how fast." That makes it a test of your SOC or MDR provider as much as your systems, and the best time to run one is after a new detection service is in place. If you are still choosing that service, the MDR vs EDR vs XDR comparison comes first.

The findings that recur are rarely exotic. In a two-week CISA assessment of a large healthcare organization, testers reached domain administrator through weak passwords, default credentials on a printer and other web interfaces, and an unpatched server, even though phishing payloads were blocked by the organization's defenses (CISA AA23-349A). That is why internal testing is worth its price even when the perimeter tests clean.

Choosing Penetration Testing Services

Penetration testing services differ less in price than in who does the work and what you get back. These are the checks that separate quotes:

Named testers

Who will test, their certifications (OSCP, CREST, GIAC and similar) and experience with your test type. Ask whether any work is subcontracted or offshore.

Manual versus automated

The share of tester-days spent on manual exploitation. A scan with a cover letter is not a pen test, and an auditor will know the difference.

The report

A sample report, with an executive summary, reproducible steps, risk ratings and fix guidance, plus an attestation letter you can hand customers.

Insurance and terms

Professional liability coverage, rules of engagement, data handling and how findings are stored and destroyed after the engagement.

HOW TO SCOPE A PEN TEST RFP

Ten Things to Put in the Scope

Send the same scope to every tester, in writing, so the quotes compare line for line and nobody prices a different job.

1
The driver and the deadlinePCI, NYDFS, CMMC, FedRAMP, SOC 2, a customer or an insurer, and the date the report is due.
2
Test typesExternal, internal, web, API, cloud, mobile, social engineering, AI, red team: which ones, and which are optional.
3
Network countsLive external IPs, internal sites, segments and the rough size of Active Directory.
4
Applications and APIsEach application with its user roles and authentication; each API with its endpoint count and documentation.
5
Cloud and AICloud accounts and key services; for AI, the model, the data it can retrieve and the tools it can call.
6
Box and accessBlack, gray or white box, test accounts, and whether source code or architecture documents are shared.
7
Windows and limitsTesting hours, production versus staging, and systems that must not be touched.
8
Pricing in tester-daysFixed price plus the days behind it, day rate for overruns and out-of-hours multipliers.
9
Retest and reportingRetest included and its window, report format, attestation letter and readout meeting.
10
Multi-year termsPrice for the same scope in years two and three, and for adding tests mid-year.

A pen test scope works best when it comes from a written inventory of what you run and where, the same discipline as an infrastructure RFP. If a test is part of buying or investing in a company, the scope belongs inside cybersecurity due diligence, which we run as paid consulting with fixed fees from $45,000.

Writing a pen test RFP?
Send us the scope instead and we will run it across the testers that fit, then negotiate days, rate and retests. The provider you choose pays us, so it costs you nothing.
Price My Pen Test

How We Source Penetration Testing

1

You send the scope

What must be tested, why, and by when, using the ten points above as a guide. You'll hear back within 24 hours.

2

We tell you what it should cost

The price range for each test from the sourced figures on this page, and which testers fit your test types and compliance driver.

3

We run the testers that fit

One scope goes to each, with tester-days, rates and retest terms requested in the same format, so quotes compare line for line.

4

We negotiate the one you choose

Days, day rate, retests, program pricing across tests and multi-year terms. The provider you choose pays us, so it costs you nothing.

Testing rarely stands alone. Findings feed the managed firewall rules and SASE policy that close external exposure, the detection service that should have caught the tester, and the backup and DRaaS plans that limit the damage when something gets through. Teams reworking network, security and hosting together can bring us in on a paid basis through IT infrastructure consulting, with engagements from $45,000.

Frequently Asked Questions

How much does a penetration test cost?

A single penetration test usually costs $5,000 to $50,000, depending on the type of test and how much is in scope. eSecurity Planet's average of 10 sources puts network tests at $9,900 to $53,700 and application tests at $8,900 to $34,600. Red team engagements run $40,000 to $200,000 or more.

How much does penetration testing cost per hour?

The median of more than 400 penetration tester rates on US federal GSA schedules is about $157 an hour, with the middle half between about $126 and $195 (MCA calculation from GSA CALC). UK G-Cloud price documents show £900 to £1,500 per tester-day.

How much does an external or internal network penetration test cost?

External network tests typically cost $5,000 to $20,000, rising toward $30,000 past about 100 live IPs. Internal network tests typically cost $7,000 to $35,000, priced by sites, network segments and the size of the directory. Each usually takes about a week of testing.

How much does a web application penetration test cost?

A web application test typically costs $5,000 to $30,000 per application, driven by user roles, functions and authentication. Vendr estimates $8,000 to $25,000 for a standard web application test on a credit-based testing platform.

How much does AI penetration testing cost?

AI and LLM penetration testing typically costs $6,000 for a single chatbot without tools to $50,000 or more for an agent system with retrieval, tools and compliance reporting, over 5 to 25 tester-days. One UK G-Cloud supplier prices AI testing at £1,600 a day against £1,200 for its standard testing.

What is LLM penetration testing?

LLM penetration testing attacks an application built on a large language model to find prompt injection, data leakage, system prompt exposure, unsafe tool use and the other risks in the OWASP Top 10 for LLM Applications, alongside the usual web, API and cloud tests of the application around the model.

How much do red team services cost?

Red team services typically cost $40,000 to $200,000 or more, with Vendr placing large-firm engagements at $75,000 to $200,000 or more. UK G-Cloud lists red team days at £1,800 against £1,200 for standard testing at the same supplier. Duration and objectives set the price.

How often is penetration testing required?

PCI DSS, NYDFS, CMMC Level 3 and FedRAMP all require testing at least every 12 months, and PCI and CMMC also after significant changes. PCI requires segmentation testing every six months for service providers. The proposed HIPAA Security Rule update would require a test every 12 months.

Does SOC 2 require a penetration test?

No. The SOC 2 criteria do not explicitly require penetration testing, but a test is widely used as evidence for monitoring criteria such as CC4.1, and many customers and auditors expect an annual report.

What is the difference between a vulnerability scan and a penetration test?

A vulnerability scan is an automated check for known weaknesses, run continuously or quarterly at low cost. A penetration test uses skilled testers to exploit weaknesses in a defined scope and show real impact, and it is what PCI DSS, NYDFS, CMMC Level 3 and FedRAMP require.

Is a retest included in the price?

Often, but not always. Some suppliers include a retest of high and critical findings, for example free up to one day's effort in one G-Cloud price document. Ask for the retest, its window and its price in writing before you sign.

What should a penetration testing RFP include?

The compliance driver and deadline, the test types, network counts, each application and API with roles and endpoints, cloud accounts and AI systems, black, gray or white box access, testing windows, pricing in tester-days with overrun rates, retest and reporting terms, and multi-year pricing.

Send the Scope Before You Request Quotes

If you have a test due for PCI, NYDFS, CMMC, SOC 2 or a customer, or you are testing an AI system for the first time, send what needs testing and when the report is due. You'll hear back within 24 hours. From there, we work out what the testing should cost and which testers fit. The provider you choose pays us, so it costs you nothing. If it isn't a fit, we will say so.

Free Pen Test Scoping Review
One scope, the testers that fit, and a negotiated price on the one you choose. Send it before any testing firm has priced you directly.
Price My Pen Test