Creyos, reviewed by a company that sells a competing test
Creyos, formerly Cambridge Brain Sciences, is a healthcare workflow product built on a 12 task cognitive battery with a real research pedigree. It reports task level scores, not a normed intelligence composite, and it does not publish a price. Here is what it fits, what it does not, and where a per participant comparison is honest.
Creyos sells cognitive tasks wrapped in clinical workflow, which is a different product from a normed intelligence composite.
0 The short answer on Creyos
Read this disclosure firstACIS sells a competing cognitive assessment, and the company behind ACIS wrote this page. There is a commercial interest in what you conclude, so weigh the review accordingly. Every claim about Creyos links to a Creyos page or a peer reviewed source, and every price carries the date it was checked. If a paragraph reads like it is steering you, follow the link.
Creyos is a clinical workflow product built around a battery of 12 short computerized cognitive tasks, and for a healthcare buyer it is a reasonable purchase that ACIS does not compete with. The tasks trace to work by Adrian Owen and colleagues at Cambridge and later at Western University in Ontario. The company sells them wrapped in condition specific protocols, automated reports, questionnaire scoring and integration with electronic health record systems. If your problem is getting cognitive data into a patient chart with a billing pathway attached, that is what Creyos is for, and this review will tell you to buy it.
The complication is that Creyos and a normed intelligence battery do not measure the same thing or report it the same way. Creyos groups its tasks into four named categories and compares performance against a normative database that Creyos states holds more than 85,000 participants. It does not publish a Full Scale IQ style composite, and it does not claim one, which means the interpretive routing set out in the guide to reading a standardized cognitive score does not apply to its output at all. It also does not publish a price on its own website. Both are defensible product choices, and both change who the platform suits.
The genuine overlap is narrower than a category page suggests. It is researchers and educators who need repeated cognitive measurement across a defined group of participants, with a cost they can forecast per person and a score they can export and analyze. That comparison is fair, it is run in detail further down, and it turns on three things: what each instrument measures, whether the output is a normed composite or a set of task scores, and what a single participant costs.
Not published
Creyos list price on creyos.com. The site routes every pricing question to a demo request. Checked September 3, 2026.
$250 per month
Starting price shown on the Capterra and GetApp directory entries for Creyos Health. Both checked September 3, 2026.
12 tasks
Cognitive tasks in the battery, grouped into four categories on the Creyos task page. Checked September 3, 2026.
1 What Creyos is, and why the old name still matters
Creyos is the current name of the company and platform that operated as Cambridge Brain Sciences until 2022, and the older name is still the one that appears in most of the published literature. The Creyos company page describes a Cambridge University team, including Adrian Owen, adapting traditional cognitive assessments into digital tests beginning around 1980, the founding of the company in 2015 when Owen connected with business leader Marc Lipton, a move to the Toronto financial district in 2021, and the rebranding from Cambridge Brain Sciences to Creyos in 2022. Owen is described as Chief Scientific Officer.
That naming history is practical, not trivia. If you search a database for validation evidence under "Creyos" you will find studies from roughly 2023 onward. Search "Cambridge Brain Sciences" or the abbreviation CBS and a much longer trail appears, including the feasibility and reliability work discussed later on this page. A buyer doing due diligence who searches only the current name will underestimate the evidence base badly, and a buyer who searches only the old name will miss the newer clinical positioning entirely.
The platform now sells under two headings. Creyos Health is the clinical product, aimed at senior focused primary care, health systems, neurology, psychiatry, occupational therapy, concussion clinics and payers. Creyos Research is the academic and commercial research product, described as a cloud based platform for collecting cognitive data at scale, with a dedicated account manager, staff scientist support and assistance with grant applications including pilot data and letters of support. Both run the same underlying tasks.
Two things follow from that structure. First, the clinical line is where the company's product investment obviously sits: the protocols, the reporting, the questionnaires and the chart integrations all serve a healthcare workflow. Second, the research line is a lighter wrapper on the same battery, which is why a research buyer should evaluate the tasks and the data handling rather than the marketing built for clinicians. Anyone weighing a supervised evaluation against any online battery should start from the tradeoffs in the comparison of proctored and unsupervised administration, and anyone unsure what the word "professional" buys in this market should read what actually distinguishes a professional grade instrument. Both apply to Creyos exactly as they apply to everything else in the category.
2 The battery: 12 tasks and the four categories they roll up to
The Creyos battery is 12 short tasks grouped into four categories, and the categories are product labels rather than a published factor structure. The Creyos task page lists them by group. Short Term Memory holds Number Ladder, Spatial Span, Paired Associates and Token Search. Reasoning holds Odd One Out, Spatial Planning, Rotations and Polygons. Concentration holds Double Trouble and Feature Match. Verbal Ability holds Digit Span and Grammatical Reasoning. The page gives roughly 2 to 3 minutes per task and describes a full battery of under 40 minutes.
Task durations differ by which Creyos page you read, and the difference is a difference in protocol rather than a contradiction. The researcher page describes a full assessment of 35 to 45 minutes. The Creyos comparison against the Montreal Cognitive Assessment describes an assessment of 5 to 15 minutes, dynamic based on the depth of assessment needed. Clinical protocols administer a subset. A researcher planning session length should budget from the full battery figure and confirm the exact task list at the demo, because the two numbers describe different products. For scale, a 35 to 45 minute battery is the same order as a supervised Full Scale administration, which Pearson states at 45 minutes for the seven WAIS-5 subtests, one of the published durations collected at the comparison of administration times across instruments, so a Creyos session is short in tasks rather than short in minutes.
Two gaps in the coverage are worth naming. Nothing in the published Creyos battery is a Gq quantitative measure of the kind ACIS runs as a timed mental arithmetic subtest, and Gs processing speed is not broken out as its own reported index. So a Creyos profile cannot answer questions about numerical reasoning or speed of simple cognitive operations, which is fine if you were not asking them. Two to three minutes per task is also short by design, the right call for a clinic slot but a measurement constraint: fewer indicators per construct means a wider band around any single score. How subtest formats differ in what they can measure bears directly on how far four category labels can be stretched.
3 The research pedigree, and what the flagship paper actually argued
The pedigree is real, and the single most cited paper behind these tasks argues against summarizing them in one number. Adam Hampshire, Roger Highfield, Beth Parkin and Adrian Owen published Fractionating Human Intelligence in Neuron in 2012, volume 76, issue 6, pages 1225 to 1237. It combined a large web based sample completing 12 cognitive tests with functional imaging, and compared factor models of individual differences against factor models of brain organization. The authors concluded that different components of intelligence have their analogs in distinct brain networks, that the higher order factor g is accounted for by tasks recruiting multiple networks at once, and that intelligence is an emergent property of anatomically distinct cognitive systems, each with its own capacity.
Read that alongside the product and it explains a design decision. Creyos reports category level scores rather than a single intelligence quotient, and that is consistent with its own scientific foundation rather than a gap in it. A buyer who wants a normed composite is not being denied one by oversight. They are looking at a platform whose intellectual starting point was the claim that one number hides more than it shows. Whether you accept that argument is a live question in the field, and the evidence for and against a general factor lays out both sides, while the account of what a cognitive test is actually measuring covers what a composite is meant to represent in the first place.
The volume claim needs a careful reading. The Creyos research studies page states that the tasks have been used in over 400 peer reviewed research studies, many in high profile journals. That is a usage count, and usage is not validation. A study that administers a battery in order to measure something else is evidence that the battery is usable and that researchers trust it enough to run it. It is not, by itself, evidence about the reliability or the criterion validity of any score the platform reports. The distinction matters when a vendor's strongest number is a count of papers rather than a coefficient.
None of this diminishes the pedigree, which is better than almost anything else in the computerized assessment market. It reframes what the pedigree buys you. It buys tasks with a long research history, a normative reference the company describes as drawn from more than 85,000 participants compiled from 14 million test scores, and a scientific officer whose name is on the foundational paper. It does not, on its own, buy you a documented standard error for the score in front of your participant. The distinction between a reliable score and a valid interpretation is the frame that turns a pedigree claim into something a buyer can check.
4 What the published evidence shows about these tasks
Three published studies give a fair picture: the battery is feasible unsupervised, its composites are reliable, and its single task scores vary a great deal in reliability. Kimia Honarmand, Conor Wild, Adrian Owen, Marat Slessarev and colleagues reported a feasibility study in PLoS One in 2019, volume 14, issue 4, article e0215203, open access at the PubMed Central full text. Twenty non delirious intensive care patients who had been mechanically ventilated for at least 24 hours completed the Cambridge Brain Sciences battery of 12 tests. Seventeen of the 20 finished, mean testing time was 45.5 minutes with a standard deviation of 11.1, and the authors identified testing related fatigue from battery length as a feasibility issue. Three patients stopped early.
The most useful psychometric study is Nicole Kochan and colleagues at the Centre for Healthy Brain Ageing in Sydney, published in the Journal of Alzheimer's Disease in 2022, volume 90, issue 4, pages 1629 to 1645, indexed at PubMed record 36314208. Fifty two community living adults with a mean age of 65.8 completed a brief battery drawn from Cambridge Brain Sciences and the Cogstate Brief Battery, unsupervised from home, on three occasions, and also sat an in person paper and pencil battery. Test retest reliability for individual measures ranged from an intraclass correlation of 0.20 to 0.83. Global cognition composites reached intraclass correlations above 0.8 over one month. A combination of computerized measures correlated with the in person battery at a canonical correlation of 0.87.
The number that should change your study designAn intraclass correlation of 0.20 on a single task means most of the variance between one sitting and the next is not the person. Kochan and colleagues found individual measures spread that widely while global composites held above 0.8. If you plan to interpret one task score for one participant, you are working at the weak end of that range. If you plan to interpret a composite across a group, you are at the strong end. That holds for Creyos, for ACIS, and for every short computerized battery.
The same paper is candid about the participant experience. Most tests were rated enjoyable, but 17 percent reported difficulty concentrating, 10 percent struggled with the computer, and 38 percent experienced performance anxiety. That last figure matters for anyone measuring memory or speed, because anxiety loads onto exactly the constructs described in the review of how memory relates to measured ability and the account of what processing speed tasks capture. A third study, Heather Sandison and colleagues in the same journal in 2023, volume 94, issue 3, pages 993 to 1004, at the open access record of that lifestyle intervention, used the battery as an outcome measure and reported gains in memory and the overall composite. The authors state plainly that their sample was small, uncontrolled, non randomized and limited to one site.
5 The criterion validity question is open, and a five year trial is running now
The study that would answer the hardest question about Creyos started in 2025 and does not report until 2029. The Centre for Addiction and Mental Health in Toronto, with the Krembil Foundation, is running an observational study registered as NCT06780917, titled Validating the On-line Creyos Cognitive Assessment Platform in Older Adults With Major Depressive Disorder or Mild Cognitive Impairment and given the short name Creyos-MD. It plans 150 participants aged 60 and older across three Toronto sites, started on January 13, 2025, and lists a primary completion date of January 2029. Participants complete in person neuropsychological testing yearly and the Creyos battery quarterly for up to five years.
Its stated primary outcomes are the ones a buyer cares about. Feasibility is measured with the System Usability Scale and the User Burden Scale after every administration. Validity is measured by comparing single test scores, domain composite scores and global composite scores against the in person paper and pencil battery, which the registration itself calls the gold standard. That is the right design, run by a group with no obvious incentive to flatter the result, and it has not reported. Anyone telling you today that Creyos is validated against gold standard neuropsychological testing in these populations is describing a study that is still recruiting.
Symmetry requires the same statement about ACIS. ACIS does not publish criterion validity coefficients pairing its Full Scale IQ against an individually administered clinical battery, because that data has not been collected at a usable sample size. What it does publish, in the technical manual documenting its norming and factor structure, is internal structure and reliability evidence on 2,750 complete records: a Full Scale IQ composite omega of .9886, a g loading of .958, a standard error of measurement of about 1.60 IQ points, and a higher order g model with CFI .9761, TLI .9726, RMSEA .0406, SRMR .0217 and a chi-square of 916.703 on 166 degrees of freedom. Those are structure figures, not criterion figures. The sample is self selected rather than census based, administration is unsupervised, and the adult reference frame of 3,243 English speaking records aged 16 to 90 is modelled rather than stratified.
So the honest scoreboard here is: Creyos has a five year criterion study in progress and a long usage record, ACIS has published internal structure and reliability figures and no criterion study, and neither has what a clinician would want before making a diagnostic decision. That is why this page never recommends an online battery in place of a clinical evaluation, and why the gold standard comparison still runs through individually administered instruments like the one described in the walkthrough of Stanford Binet 5 administration. The evidence on how accurate unsupervised online testing actually is sets the ceiling that applies to both products.
6 Creyos is a healthcare workflow product, and for most buyers that settles it
If you need cognitive data inside a chart with a billing pathway, buy Creyos and stop comparing it to anything ACIS sells. This is not a hedge. ACIS is not a clinical instrument, is not proctored, is not approved for diagnosis, and does not integrate with any electronic health record. Those are stated boundaries of the product, not gaps waiting to be filled. A clinician evaluating both is comparing a purpose built healthcare tool against something that was never built for the job, and the comparison should end there.
What Creyos sells on that side is workflow, not psychometrics. Its comparison page against the MoCA names direct integration with EPIC, athenahealth and eClinicalWorks, automatic scoring with domain level breakdowns rather than a single number, automatic comparison against a patient's previous results, and multilingual administration. It positions the product between a brief screener and a full neuropsychological evaluation, describing purpose built protocols with condition specific focus. That positioning is accurate to what the product does.
The commercial argument is reimbursement. The Creyos reimbursement calculator claims providers may be eligible for more than $150 of reimbursement per assessment through public and private insurers, naming CPT codes 96132 and 96138. Read the disclaimer on the same page, which assumes a 60 percent reimbursement success rate and states that Creyos cannot determine your eligibility and does not assume responsibility for the outcome of any claims. That is an appropriately worded caveat from the vendor, and it means the reimbursement case is yours to verify against your own payer mix rather than a figure to drop into a budget.
Where this review sends you elsewhereIf you are a licensed clinician conducting a formal cognitive evaluation, neither Creyos nor ACIS replaces an individually administered instrument delivered by a qualified examiner, and neither claims to. A short computerized battery answers a different question than a full protocol with behavioral observation, and the index structure and administration demands of the WAIS-5 show what that difference costs in examiner time.
One workflow point matters for research buyers reading the clinical marketing. Creyos does not publish how it computes or reports a score on the pages reviewed for this article. The task page states that a patient's result is compared to a normative database of over 85,000 subjects and tabulated instantly into an easy to interpret report, without specifying whether the reported figure is a percentile, a standardized score, or something else. Ask for a sample report and a scoring specification at the demo. A report you cannot reproduce from underlying data is not a defensible outcome measure, and the sequence a buyer should expect from any adult assessment is laid out in the step by step account of how an adult cognitive assessment runs.
7 The ADHD protocol, and what a cognitive battery can settle
The Creyos ADHD protocol pairs five cognitive tasks with a symptom questionnaire and takes under 25 minutes, and Creyos positions it as an aid to clinical judgment rather than a diagnostic test. The Creyos ADHD assessment page names the five tasks as Spatial Planning for planning, Token Search for spatial working memory, Feature Match for attention, Double Trouble for response inhibition, and the Sustained Attention to Response Task for sustained attention. It states the assessment measures 14 validated markers of ADHD and pairs them with an age appropriate questionnaire: the Vanderbilt scale for children aged 6 to 12, the SWAN across the full spectrum, and the Adult ADHD Self-Report Scale Part A for adults.
The framing on that page is careful, and the care is worth crediting. Creyos describes the protocol as helping clinicians objectively evaluate ADHD symptoms, as data that complements rating scales and DSM-5 criteria, and as support for confident clinical decisions. It does not claim that the protocol diagnoses ADHD on its own. That is the correct claim to make, because no computerized cognitive battery in existence diagnoses ADHD. Performance on attention and inhibition tasks overlaps heavily between diagnosed and undiagnosed groups, and a group difference in a published study is never an individual diagnosis.
For a reader who arrived searching for a Creyos ADHD test, the practical answer has three parts. First, if you already conduct ADHD evaluations, the protocol is a structured way to add objective task data and a scored questionnaire to a workflow you run anyway, and the 25 minute figure is realistic for a clinic slot. Second, if you are an adult wondering whether you have ADHD, no online battery answers that, and the route is a clinician rather than a purchase. Third, if you are a researcher studying attention, the five task subset is a plausible measurement package, but you will want item level data and the scoring rules before committing.
ACIS has nothing to offer on this question and says so. It is not built for ADHD screening, its results are explicitly not intended for accommodation determinations, and it produces no attention deficit index of any kind. Its working memory and processing speed indices measure Gwm and Gs as cognitive constructs, which is a different target from a clinical attention profile. If your interest is the research question rather than the diagnostic one, the published findings on how ADHD relates to measured cognitive ability covers the group level evidence and the limits on reading it individually, and the wider survey of mental health and cognitive test performance explains why state factors move scores in ways a single sitting cannot separate.
8 What Creyos costs, and what is not published
Creyos does not publish a price on its own website, and the only public figures come from third party software directories. The direct route was checked first: creyos.com/pricing returned a 404 Not Found on September 3, 2026, and every pricing pathway on the site resolves to a demo request form. That is a common enterprise sales choice and it is not evidence of anything beyond itself. The price is not published, and a review should say that rather than guess at a number.
Two directory listings carry a figure. The Capterra listing for Creyos Health, which still sits at a URL carrying the legacy CBS Health identifier, showed a starting price of $250.00 per month, a free trial available and no free version, when checked on September 3, 2026. It also showed no user reviews at all on that date, so there is no aggregate rating there to read. The GetApp pricing page for Creyos Health showed the same $250 per month starting figure with a stated last update of August 2026, also checked on September 3, 2026, but marked both a free trial and a free version as available. The two listings disagree on the free version, which is a reminder that directory entries are compiled summaries and not vendor price lists.
A listed starting price is not a quoteEvery arithmetic result in the next section rests on the $250 per month figure shown on two directories, treated as a floor, with no per assessment charge assumed on top. Creyos has not confirmed that figure publicly, the real quote depends on practice size and contract term, and a starting price by definition describes the cheapest configuration rather than yours. Get a written quote before planning a budget around any number on this page.
The structural consequence matters more than the figure. A monthly subscription floor means cost per participant is entirely a function of volume, and you cannot compute it before a sales call. At a $250 monthly floor, five participants in a month costs $50 each, twenty costs $12.50 each, and a hundred costs $2.50 each. Run a study with a slow recruitment tail and the fixed monthly cost keeps accruing while the participant count does not. Run a high volume clinic and the per patient cost falls to almost nothing. Subscription pricing rewards steady volume and punishes sparse, bursty use.
That is the opposite of how a credit model behaves, which is why the next section runs the comparison rather than asserting a winner. Neither model is better in the abstract: one is cheaper for a clinic seeing patients weekly, the other for a team that runs 30 participants in March and nothing until September. For the wider landscape, the breakdown of what cognitive testing costs across formats puts both next to a proctored clinical evaluation. The organizational side of the same question, where Pearson prints $1.95 for each comprehensive subtest administered and $310 per examiner seat per year, is worked out at the per administration arithmetic of testing other people, and a subscription floor is the hardest shape to place inside that table.
9 Cost per participant, the ground where the two products overlap
On a per participant basis the two products invert each other: Creyos gets cheaper the more you run, ACIS Professional costs the same per person whether you run three or three hundred. ACIS publishes its prices, so that column is a list price rather than an estimate. A Professional account is free to open and includes three Quick administrations at no cost. After that one credit equals one US dollar: Quick costs 15 credits, Optimized 30, Full Scale 50, and a custom battery 5 per subtest. Credits do not expire and no subscription is required, as set out on the ACIS research workspace page.
Cost question
Creyos
ACIS Professional
Price published by the vendor
No. creyos.com/pricing returned 404 on September 3, 2026 and all routes lead to a demo request.
Yes. Per administration prices are listed publicly.
Public figure available
$250 per month starting price on Capterra and GetApp, both checked September 3, 2026.
15, 30 or 50 credits per administration, 1 credit equals $1.
Pricing model
Subscription with a monthly floor.
Pay as you go credits, with optional monthly plans.
First three participants
At least one month at the listed floor, so roughly $250 on that assumption.
$0. Three Quick administrations are included with a free account.
20 participants in one month
Roughly $250 at the listed floor if no per assessment charge applies, about $12.50 each.
$255 at the Quick tier after the three included, about $12.75 each.
20 participants over six months
Roughly $1,500 at the listed floor, about $75 each, because the floor accrues monthly.
$255. Credits do not expire and no month passes unused.
200 participants in one month
Roughly $250 at the listed floor, about $1.25 each, subject to the actual contract.
$2,955 at the Quick tier, about $14.78 each.
Commitment before you see a price
A sales conversation.
None. Registration takes an email and a password.
The crossover is obvious once the numbers sit side by side. Below roughly twenty participants per month the two land within a few dollars of each other, and ACIS is cheaper for anything sparse because nothing accrues while you are not testing. Above that, Creyos wins on unit cost and the gap widens fast, exactly the shape you would expect from a product built for a clinic that tests every week. A researcher running one cohort a year sits on the ACIS side. A health system screening hundreds of patients a month sits on the Creyos side.
Two caveats sit on the Creyos column. Every figure in it rests on a directory listed floor rather than a quote, and the real contract may carry per assessment charges, seat limits or annual terms that change the arithmetic entirely. Ask whether the subscription includes unlimited administrations, and get the answer in writing. The ACIS caveat differs in kind: prices are certain, but administration is unsupervised, which constrains study design rather than budget. Length is the other variable, since the 20 subtest form described in the guide to full length cognitive batteries asks far more of a participant than a 6 subtest short form.
10 What each one measures and how it reports a score
The reporting difference is larger than the measurement difference, and it is the thing most buyers get wrong. Creyos gives you task level performance in four categories, benchmarked against its normative database. ACIS gives you scaled scores on 20 subtests, six index scores and a Full Scale IQ on the conventional metric. Both are legitimate outputs answering different questions, and picking the wrong one means collecting data you cannot use.
Dimension
Creyos
ACIS
Number of measures
12 tasks.
20 subtests, with 13 subtest and 6 subtest short forms.
Grouping
Four categories: Short Term Memory, Reasoning, Concentration, Verbal Ability.
Six CHC domains reported as VCI, FRI, QRI, VSI, WMI and PSI.
Composite reported
No intelligence quotient style composite is published or claimed.
Full Scale IQ, plus six primary indices.
Score metric
Not specified on the public task page beyond comparison to the normative database.
Scaled scores at mean 10 and standard deviation 3, composites at mean 100 and standard deviation 15.
Reference sample
Creyos states more than 85,000 normative participants drawn from 14 million test scores.
An adult reference frame of 3,243 English speaking records, ages 16 to 90, self selected rather than census based.
Published reliability
Not published on creyos.com as of September 3, 2026. Peer reviewed estimates exist for the tasks, cited above.
Full Scale IQ composite omega of .9886 and a standard error of measurement of about 1.60 IQ points, on N = 2,750 complete records.
Published factor evidence
The 2012 Neuron paper argues for multiple components rather than one.
Higher order g model with CFI .9761, TLI .9726, RMSEA .0406, SRMR .0217, chi-square 916.703 on 166 degrees of freedom, and a g loading of .958.
Supervision
Clinic or at home, depending on protocol.
Unsupervised throughout. A stated limit, not a feature.
Data export
Not specified publicly. Ask for the export format at the demo.
A ZIP containing cohort_summary.csv, scores_long.csv and codebook.csv.
Every ACIS figure above carries the same three limitations, and they travel with it. The normative sample is self selected rather than census based, administration is unsupervised, and the adult reference frame is modelled rather than stratified. High internal consistency on a self selected sample tells you the items hang together. It does not tell you the reference population resembles yours, which is the whole burden of the explanation of how a norm table is actually constructed. Why composites sit on a 15 point standard deviation covers the metric that makes those numbers comparable across instruments at all.
The practical routing follows from the composite row. If your analysis needs a single index of general cognitive ability that maps onto a literature full of IQ points and effect sizes on that metric, you need a normed composite, and Creyos does not publish one. If it needs task profiles across attention, memory and reasoning tracked over repeated sittings in a clinical population, the composite is beside the point. What a Full Scale IQ actually summarizes, the score bands it is read against and the percentile each point converts to cover the interpretive side of that choice.
11 Who should buy which, stated plainly
Most people reading a Creyos review are not choosing between Creyos and ACIS, and the honest answer for several of them is neither. The table below routes by buyer, including to third parties and to no purchase at all. It is written to be usable by someone who has already decided ACIS is an interested party on this question, which is a reasonable thing to have decided.
Who you are
What to do
Why
Clinician wanting cognitive data in the patient chart
Creyos.
Chart integration, condition protocols and a reimbursement pathway are the product. ACIS has none of them and does not claim to.
Neuropsychologist conducting a formal evaluation
Neither. Use an individually administered clinical instrument.
No short computerized battery substitutes for a full protocol with behavioral observation by a qualified examiner.
Academic researcher needing a normed general ability composite
ACIS Professional, or a licensed clinical battery if the budget allows.
Creyos does not publish an intelligence quotient style composite. If your model needs one, task scores will not substitute.
Researcher tracking cognitive change in a clinical population
Creyos.
Repeated administration, automatic comparison to prior results, and a five year validation study under way in exactly that population.
Educator or program evaluator with a small cohort
ACIS Professional.
Three free administrations, a published per participant price, no subscription floor, and CSV export.
Employer screening job candidates
Neither.
Neither product is validated for employment decisions, and both exclude that use. Job relevant validation is a separate discipline.
Adult who wants to know their own score
Neither of these. Read the consumer options.
Creyos sells to organizations only. ACIS sells a consumer test, which is a different page and a different argument.
The employer row deserves an extra sentence, because it is where the most money gets wasted. Cognitive ability does predict job performance in the aggregate literature, but a general cognitive score is not a validated selection instrument on its own, and using one without job analysis and local validation invites both bad hiring and legal exposure. What a legally defensible pre-employment cognitive test requires and the actual size of the ability to performance correlation cover the distance between a correlation and a hiring tool.
12 The questions to put to any cognitive assessment vendor
The fastest way to evaluate a platform is to ask nine questions and watch which ones get answered with a document rather than a sentence. The checklist applies to Creyos, to ACIS and to every competitor in the category, and every good answer on it is checkable after the call. Which competitors belong in the comparison depends on the category you are actually buying in, since the phrase cognitive assessment platform covers clinical screeners, research task builders, consumer brain training and normed ability batteries, and only the last of the four returns a composite on a population scale, a split mapped at the survey of the four platform categories.
Ask this
Why it matters
What a good answer looks like
Is there a technical manual, and may I have it before I buy?
A manual is where norming, reliability and factor evidence live. A brochure is not a manual.
A document you can read now, not one promised after signature.
What is the reliability coefficient, and for which score?
Composite and single task reliability differ enormously, as Kochan and colleagues showed.
A coefficient attached to a named score and a named sample.
What is the standard error of measurement in reported units?
Without it you cannot say whether a change between two sittings means anything.
A figure in score units, with the confidence interval it implies.
How was the norm sample recruited?
A self selected database and a stratified population sample are not interchangeable.
An honest recruitment description, self selected included if true.
Is administration supervised, and what follows if it is not?
Unsupervised testing changes the interpretation of every score, in both products here.
A stated position on what unsupervised results can support.
What exactly can I export, and in what format?
A report you cannot reproduce from underlying data is not a research outcome.
Named files or fields, and a sample export you can open.
What is the retest interval, and are there alternate forms?
Practice effects contaminate repeated measurement without a plan for them.
A documented interval and a statement about form equivalence.
What participant data is stored, and where?
Institutional review boards ask this before approving anything.
A data processing description you can attach to an ethics application.
What does one participant cost at my actual volume?
Subscription floors and credit models give very different answers.
A written quote with the volume assumption stated.
Apply the checklist to this page's own subject and the pattern is clear. Creyos answers the workflow, data handling and export questions well in a sales conversation, and publishes no technical manual, no reliability coefficient and no price. ACIS publishes all three and still cannot satisfy a clinical buyer on supervision, because administration is unsupervised. Both gaps are real, and neither is hidden.
Two questions carry more weight than their length suggests. Recruitment decides whether a percentile means anything for your participants, and the evidence on measurement bias across groups explains why a large convenience sample does not settle it. The multilingual claim deserves the same scrutiny, since translating instructions is not the same as demonstrating equivalence, a distinction developed in the account of what culture reduced testing can achieve. Add one habit: search the literature under every name a product has had, because a rebranded platform files most of its evidence under the old one. Where nothing exists under any name, the gap between an unvalidated quiz and a documented instrument is the finding.
13 The verdict, and the standard both products should be held to
Creyos is a well built healthcare product with a genuine research pedigree, an undisclosed price, and no published intelligence composite, and those three facts route almost every buyer on their own. The tasks descend from work by Adrian Owen and colleagues, and the 2012 Neuron paper remains one of the more interesting arguments in the field about whether general ability should be reported as one number at all. The clinical wrapper around those tasks is what a healthcare organization is actually buying, and it is the part ACIS does not compete with. For a researcher or educator who needs a normed composite at a forecastable per participant cost, the fit is worse.
The complaints on this page are attributed, and there are fewer of them than a competitor review usually contains. Honarmand and colleagues in 2019 identified fatigue from battery length, with three of twenty patients stopping early. Kochan and colleagues in 2022 found single measure test retest reliabilities as low as an intraclass correlation of 0.20 alongside composites above 0.8, and reported that 38 percent of their older adult participants experienced performance anxiety during unsupervised home testing. Capterra listed no user reviews on September 3, 2026, so there is no aggregate customer rating to summarize. Where something is not published, this page says it is not published, which is a different statement from saying it does not exist.
The standard both products should be measured against is the same one, and neither meets all of it. The Standards for Educational and Psychological Testing, published in 2014 by the American Educational Research Association, the American Psychological Association and the National Council on Measurement in Education, require a developer to document the intended uses of a score, the evidence supporting each interpretation, the composition and recruitment of the normative sample, reliability estimates for every reported score, and the conditions of standardization. The APA testing standards make the same demand in the same spirit: a score is only as defensible as the documentation that travels with it. Instruments that clear that bar publish it openly, which is why the published manual behind Raven's 2 is a useful yardstick.
Measured that way, Creyos publishes strong evidence of use and a large normative claim, and does not publish reliability coefficients, standard errors or a price on its own site. ACIS publishes reliability and structure figures, prices and export formats, and cannot offer supervised administration, a census based norm sample, or criterion validity evidence against a clinical battery. A buyer should require the missing documents from whichever vendor they choose rather than accept marketing in their place, and the criteria that separate an accurate cognitive test from a plausible looking one is a workable list to hold them to. That request is what the Standards exist to license, and any vendor who treats it as unreasonable has answered it.
14 Frequently Asked Questions
What is Creyos?
Creyos is a cognitive assessment platform sold to healthcare organizations and research groups, built on twelve short computerized tasks. It ships as Creyos Health for clinics and Creyos Research for studies.
Is Creyos the same company as Cambridge Brain Sciences?
Yes. The Creyos company page describes rebranding from Cambridge Brain Sciences in 2022, and the older name is the one attached to most of the published literature on these tasks.
How much does Creyos cost?
Creyos publishes no price, and creyos.com/pricing returned a 404 when checked on September 3, 2026. Capterra and GetApp both listed a starting price of $250 per month that day.
Why does Creyos not show its pricing?
Nothing suggests concealment. Routing pricing through a sales call is standard enterprise practice, so the accurate description is that the price is not published, not that it is hidden.
Does Creyos give you an IQ score?
No. Creyos reports task performance in four categories against its normative database, and neither publishes nor claims an intelligence quotient style composite. If your analysis needs one, look elsewhere.
Who is Adrian Owen and why does his name appear everywhere?
Owen is a neuroscientist at Western University and Chief Scientific Officer at Creyos. The platform's tasks descend from his research group, and he co-authored the 2012 Neuron paper Creyos cites most.
How many cognitive tasks does Creyos include?
Twelve. The Creyos task page names Number Ladder, Spatial Span, Paired Associates, Token Search, Odd One Out, Spatial Planning, Rotations, Polygons, Double Trouble, Feature Match, Digit Span and Grammatical Reasoning.
How long does the full Creyos battery take?
Creyos gives two to three minutes per task and a full battery under forty minutes, while its researcher page says thirty five to forty five. Clinical protocols run a shorter subset.
Is Creyos validated?
The tasks have a long publication record and Creyos states they appear in over 400 peer reviewed studies. A usage count is not a validation coefficient, and none is published on creyos.com.
Is anyone testing Creyos against gold standard neuropsychological testing?
Yes. Study NCT06780917, run by the Centre for Addiction and Mental Health, compares the battery against in person testing in 150 older adults, with primary completion listed for January 2029.
Can Creyos diagnose ADHD?
No, and Creyos does not claim it can. Its ADHD protocol is described as complementing rating scales and DSM-5 criteria to support a clinician's judgment, not as a diagnostic test.
What is in the Creyos ADHD protocol?
Five tasks: Spatial Planning, Token Search, Feature Match, Double Trouble and the Sustained Attention to Response Task, paired with an age appropriate questionnaire. Creyos describes it as taking under twenty five minutes.
Does Creyos integrate with electronic health records?
Its MoCA comparison page names EPIC, athenahealth and eClinicalWorks. Confirm your own system at the demo, because other Creyos pages describe integration only in general terms.
Can I bill insurance for a Creyos assessment?
Creyos publishes a calculator claiming over $150 per assessment via CPT codes 96132 and 96138. The same page states Creyos cannot determine eligibility or take responsibility for claim outcomes.
How does the cost per participant compare with ACIS Professional?
Around twenty participants a month the two land within a few dollars of each other. Below that ACIS is cheaper because credits never expire, and well above it Creyos is cheaper.
Is ACIS a substitute for Creyos in a clinic?
No. ACIS is unsupervised, is not a clinical or diagnostic instrument, and has no chart integration or billing pathway. A clinician who needs data in the record should buy Creyos.
What does ACIS publish that Creyos does not?
Per administration prices, an export specification, and psychometric figures including a Full Scale IQ composite omega of .9886 on 2,750 complete records. That sample is self selected and unsupervised.
What does Creyos publish that ACIS does not?
A normative database described as more than 85,000 participants, a long record of use in peer reviewed studies, condition specific clinical protocols, and integration with major electronic health record systems.
Are there user reviews of Creyos worth reading?
The Capterra listing showed no user reviews at all on September 3, 2026, so there was no aggregate rating to summarize. The peer reviewed literature is more informative here.
Should I trust a Creyos review written by a competitor?
Not on faith. Every claim here links to a Creyos page or a peer reviewed source, and every price carries the date it was checked, so the load bearing parts are checkable.
What should I ask a cognitive assessment vendor before buying?
Ask for the technical manual, the reliability coefficient for the exact score you will report, the standard error in score units, how the norm sample was recruited, and a written quote at your volume.
Take the assessment
You get a profile, not a number
ACIS measures six CHC domains across 20 subtests and reports each one with its own normed score and confidence interval, so you can see where you are strong and where you are not.
