Radiological.ai

Buyer's guide · real award data

Radiology Software Pricing: Models, Comparison and What US Groups Actually Pay

The short answer

Radiology software is priced five ways: per named user, per study, per site, per clinical module, or as a negotiated enterprise agreement. Almost no vendor publishes list pricing. The clearest public benchmark is US federal contract data, where the median award for PACS is $250,438, the median award naming PowerScribe is $92,403, and the median teleradiology services award is $13,501.

This page is the pricing research a radiology group has to do before it can read a quote properly. It covers how each category is priced, what actually moves the number, and the figures US institutions have really paid.

Last updated August 2026

The Reading Station

Worklist

SERIES 1 · AX
SLICE 24/64
SAMPLE STUDY
NOT FOR DIAGNOSTIC USE
W 80 · L 40
ILLUSTRATIVE SAMPLE

Structured report

Draft

Run the assistant to draft this report for review.

You review & sign

Illustrative sample · not a real patient study, not a diagnosis

Drafted in · you review & sign Worklist re-prioritized

Decision support for qualified clinicians. Radiological.ai does not provide a diagnosis and is not a substitute for professional judgment.

Jump to the award data

Flag · prioritize · draft · you review and sign

$250,438

Median US federal award for PACS and image archiving, across 48 awards

$92,403

Median US federal award naming PowerScribe, across 161 awards

$308.9M

Total across all three categories, FY2017 through August 2026

Radiology software pricing models

The five ways radiology software is priced

Vendors quote in different units on purpose, because it makes side-by-side comparison hard. Knowing which model you are being quoted in is the first step of any real comparison.

Per named user, per month

You pay for each radiologist with a login. Predictable, and it does not punish a busy month. It does punish a group that keeps seats open for locums who read twice a year.

Usually: Reporting platforms, dictation, worklist, most AI assistants.

Per study or per exam

You pay each time a study runs through the software. It scales down cleanly for a small site and becomes the most expensive model the moment volume grows. Watch for a minimum-volume floor buried in the order form.

Usually: Point AI tools, screening products, some cloud PACS.

Per site or per facility

A flat license for one location, usually with a study ceiling. Good for a single imaging center. Multi-site groups end up negotiating each new location separately, which is exactly where the price creeps.

Usually: RIS, enterprise PACS, hospital AI deployments.

Per module or per pathway

You license each clinical pathway on its own, so stroke, pulmonary embolism and fracture are three line items. It is the model that quietly turns a $60,000 quote into a $200,000 one over three renewals.

Usually: Acute triage platforms such as Aidoc, Viz.ai and RapidAI.

Enterprise agreement

A negotiated bundle across the whole system, usually multi-year, usually with an uplift clause. The list price stops mattering and the renewal terms start mattering a great deal.

Usually: Health systems, IDNs, anything above roughly 40 radiologists.

Radiology software pricing comparison

What US institutions have actually paid

Private radiology software quotes are confidential, but federal ones are public. These are prime contract awards recorded on USAspending.gov, pulled on August 19, 2026, covering award start dates from October 1, 2016 through August 19, 2026.

Category Awards Total 25th pct Median 75th pct 90th pct
PACS and image archivingsearch term: picture archiving and communication 48 $68,836,793 $78,263 $250,438 $1,046,144 $4,466,524
Reporting and dictation (PowerScribe)search term: powerscribe 161 $38,202,976 $24,813 $92,403 $223,962 $634,826
Teleradiology servicessearch term: teleradiology services 1,117 $201,926,209 $5,636 $13,501 $60,450 $473,698

PACS and image archiving. The most expensive line in the stack. Storage and enterprise scale drive it.

Reporting and dictation (PowerScribe). Bought far more often than PACS, at roughly a third of the median size.

Teleradiology services. A service, not a license. Thousands of small awards plus a few very large ones.

Read the spread rather than the median. In every category the 90th percentile is between five and thirty five times the 25th percentile, which tells you the same product is sold at wildly different sizes depending on scope. Your job in a negotiation is to establish which end of that spread your group actually sits at, before the vendor establishes it for you.

Source: USAspending.gov award search API, prime awards, award type codes A, B, C and D. Keyword matching is approximate, so figures are indicative of deal size rather than a precise product-level total.

Read this before you call the vendor

Most of the money does not go to the vendor

The single most useful thing in the federal data is not a price. It is who got paid. Of the $38.2 million in awards naming PowerScribe, about $24.7 million went to Carahsoft Technology Corp, a reseller. Only about $1.6 million, roughly 4%, was awarded directly to Nuance Communications.

That pattern holds across enterprise radiology software generally. The name on the contract is usually a reseller or a systems integrator, not the company that wrote the software. It matters for three practical reasons. Discount authority often sits with the reseller rather than the vendor, so the person who can actually move the price may not be the one in the demo. Support escalation paths run through whoever holds the paper. And renewal uplift is frequently a reseller term, not a vendor term, which is why it is so often the clause nobody can explain.

Ask early which entity will hold the contract. If the answer is a reseller, ask for the vendor's own pricing sheet as well, and ask both parties the same question about the year-three renewal.

The same discipline applies when you are evaluating AI radiology software alongside the core stack. Per-module platforms and per-seat assistants look comparable in year one and diverge sharply by year three.

Radiology software pricing factors

Six things that move the number on your quote

Bring your own figures for all six to the first call. Every one you cannot answer becomes an assumption the vendor makes in its own favor.

Reading radiologists, not headcount

Vendors quote per seat, but the number that matters is how many people actually read. Groups routinely pay for administrators, residents and inactive locums because nobody audited the seat list before signing.

Annual study volume

Per-study pricing and volume tiers both key off this. Bring your real number for the last twelve months, broken out by modality, or you will be quoted against the vendor's guess.

Modality mix

CT and MR cost more to store, more to move and more to process than plain film. A group that is 70% X-ray and a group that is 70% cross-sectional get very different quotes on identical volume.

Integration scope

One PACS over DICOM is cheap. Bidirectional HL7 or FHIR into the EHR, a second PACS from a merger, and a custom worklist feed are each a separate professional-services line.

Storage and retention

State retention rules and study size compound. This is the driver that makes PACS the biggest number in the federal data and the one most often left off the first-year estimate.

Support and maintenance

Commonly quoted by third-party software aggregators at roughly 15% to 20% of the license fee per year. On a $100,000 license that is another $15,000 to $20,000 you did not put in the board deck.

Radiology software pricing negotiation

How to make four quotes comparable

Vendors quote in the unit that flatters them. A per-study tool quotes cents. A per-module platform quotes one pathway. A per-seat platform quotes a monthly rate with the implementation left out. None of those numbers can be compared to each other as given.

Convert everything to two numbers and only two. The first is fully loaded first-year total: license, implementation, integration, training, storage, support, and any hardware. The second is year-three annual run rate, calculated with the uplift clause applied twice and with your projected volume rather than today's. Put both in a four-column table with one vendor per column.

Then ask each vendor the same five questions in writing. What is the renewal uplift cap? What happens to the price if volume rises 30%? What is the storage overage rate? Which entity holds the contract? And what does it cost to leave, including data export in a usable format? The answers you have to chase are the ones that matter.

If reporting is the line you are pricing, the honest comparison usually comes down to a per-seat platform against a per-module one. We have written that comparison out in detail on our PowerScribe pricing page, and the broader category tradeoffs live on structured radiology reporting.

Questions buyers actually ask

Radiology software pricing questions

How much does a PACS system cost?

Federal prime contract data gives the clearest public answer. Across 48 US federal awards naming picture archiving and communication from FY2017 through August 2026, the median award was $250,438 and the total was $68.8 million. A quarter of awards came in under $78,263 and a tenth exceeded $4.4 million. Private-sector quotes track the same shape: small single-site deployments in the low tens of thousands, enterprise systems in seven figures.

What are the radiology software pricing models?

There are five. Per named user per month, per study or exam, per site or facility, per clinical module or pathway, and a negotiated enterprise agreement. Reporting and AI assistants usually price per user. Point AI tools usually price per study. Acute triage platforms usually price per module, which is the model that escalates fastest as you add pathways.

How much does PowerScribe cost?

Microsoft does not publish list pricing for PowerScribe. The public federal record is the best available anchor: 161 US federal prime awards naming PowerScribe from FY2017 through August 2026 total $38,202,976, with a median award of $92,403 and a 75th percentile of $223,962. Those figures cover licenses, renewals and services together, so treat them as deal sizes rather than per-seat rates.

Why do radiology software vendors not publish pricing?

Because the number depends on volume, modality mix, seat count and integration scope, and because every deal is negotiated. Publishing a list price would anchor every negotiation against the vendor. The practical consequence for a buyer is that comparison shopping requires you to force all vendors onto the same quote template, which is the single highest-leverage thing you can do.

What is a fair price for radiology reporting software?

There is no single fair price, but there is a fair test: total three-year cost divided by studies read over those three years. Run that number for every quote on the table. It exposes the per-module platform that looks cheap in year one, and it exposes the per-study tool that looks cheap until volume grows.

What should be included in a radiology software quote?

Six things, in writing: the license itself, implementation and integration services, annual support and maintenance as a stated percentage, storage and its overage rate, training, and the renewal uplift cap. A quote missing the uplift cap is not a quote, it is a first-year discount.

How do you compare radiology software pricing across vendors?

Normalize before you compare. Convert every quote to fully loaded first-year total and to year-three annual run rate, using your own volume and seat numbers rather than the vendor's assumptions. Vendors on different pricing models are otherwise impossible to line up, which is exactly why they quote in different units.

For comparison

What we charge, published

We publish our rate because the research above is tedious and most of it exists only because nobody else will.

Radiological.ai is priced per reading radiologist, per month, not per study. Group is $299 per radiologist per month and Practice is $399, with a custom enterprise tier above that. One assistant covers flagging suspected findings, worklist prioritization and structured report drafting across X-ray, CT and MRI, so it is one line item rather than three, and there is no per-pathway module list to grow into.

Full plan detail, including what is in each tier, is on the Radiological.ai pricing page.

One line item instead of three

Radiological.ai flags suspected findings, prioritizes the worklist and drafts the structured report on X-ray, CT and MRI, priced per reading radiologist. You review and sign every study.

See plan pricing

Decision support for qualified clinicians. Radiological.ai does not provide a diagnosis and is not a substitute for professional medical judgment.