- What "Passing" Actually Means on This Exam
- Why There Is No Published Percentage
- Exam Format That Shapes Your Score
- Where Your Points Come From: The Five Weighted Domains
- Reading Your Domain-Level Feedback
- Attempts, Retakes, and the 12-Month Window
- Planning Around a Criteria-Based Standard
- Exam-Day Rules That Protect Your Result
- After You Pass: Renewal and Career Use
- Frequently Asked Questions
- PMI uses a criteria-based, psychometrically established passing standard; no numerical passing percentage has been verified for this exam.
- The exam has 120 questions, but only 100 are scored; 20 are unscored pretest items.
- Identify Business Needs and Solutions and Identify Data Needs are 26% each, together 52% of the blueprint.
- You receive pass/fail status plus domain-level diagnostic feedback, not a raw score.
What "Passing" Actually Means on This Exam
Search for the passing score of the Cognitive Project Management for AI credential and you will find plenty of confident numbers floating around. Most of them are guesses, or worse, borrowed from unrelated certifications that happen to share the same acronym. The honest answer is more nuanced and, once you understand it, more useful for planning.
The current exam is the PMI Certified Professional in Managing AI (PMI-CPMAI), introduced on September 30, 2025 to replace CPMAI v7. It is administered by the Project Management Institute (PMI), building on the methodology developed by Cognilytica. Rules from the legacy v7 exam and from CPMAI+E are not interchangeable with this exam, so older forum posts and study notes about "the passing percentage" may describe a different version entirely.
If you want the bigger picture on how hard that standard is to meet, our guide on how hard the CPMAI exam is breaks down the difficulty from a candidate's perspective, and the CPMAI pass rate analysis explains what can and cannot be said about outcomes.
Why There Is No Published Percentage
A criteria-based standard is different from a simple "get 70 percent right" rule. Here is the practical logic behind it.
Criteria-based versus fixed percentage
Professional certification bodies typically establish a passing standard through a structured process in which subject-matter experts judge how a minimally qualified candidate would perform on the content. The resulting standard reflects the difficulty of the questions on the form, not an arbitrary round number. Two exam forms can contain different questions of different difficulty, so the standard is set to represent the same level of competence across forms rather than the same raw count of correct answers.
What this means for you
- Do not chase a magic percentage. Aiming for "just over 70" on practice tests tells you little about your readiness.
- Aim for competence across the blueprint. Because the standard is tied to demonstrated knowledge, broad and consistent command of all five domains is the safest strategy.
- Treat practice scores as directional. A practice test is a rough indicator of strengths and gaps, not a prediction of your official outcome.
Exam Format That Shapes Your Score
Understanding the structure of the test helps you reason about what a passing performance looks like. These details come directly from the exam administration facts.
| Element | Detail |
|---|---|
| Total questions | 120 |
| Scored questions | 100 |
| Unscored pretest items | 20 |
| Testing time | 160 minutes |
| Scheduled breaks | None |
| Delivery | Pearson VUE test centers or OnVUE remote proctoring (subject to regional availability) |
| Prerequisite | Completion of the 21-hour PMI-CPMAI Exam Prep Course |
The 20 unscored questions
Of the 120 questions you will see, 20 are pretest items that do not count toward your result. You will not be told which ones they are. The practical consequence: treat every question as if it counts, and do not let a bizarre or unfamiliar item rattle you. It may well be one of the unscored ones being trialed for future exams.
Time pressure
You have 160 minutes for 120 questions, which works out to a little over a minute and a half per question. There are no scheduled breaks, so pacing and stamina matter. Tutorial and survey time sit outside the exam timer, so the 160 minutes is purely testing time. Note also that the 21 hours associated with the course is the length of the preparation course, not exam time. If you want a deeper plan for pacing, the CPMAI study guide covers how to build stamina before test day.
Scenario orientation
The credential is built around managing AI projects through a data-centric, iterative methodology, so expect questions that ask you to choose the best next action in a project situation rather than recite definitions. Questions tend to reward knowing why a step belongs in a particular phase, such as why data understanding must precede model development, rather than memorizing vocabulary.
Where Your Points Come From: The Five Weighted Domains
Since the passing standard is criteria-based, the single best predictor of success is competence across the weighted exam domains. The official content outline defines five domains. Exact question counts per domain can vary by exam form, so treat the weights as proportions, not guarantees.
| Domain | Weight |
|---|---|
| Domain 1: Support Responsible and Trustworthy AI Efforts | 15% |
| Domain 2: Identify Business Needs and Solutions | 26% |
| Domain 3: Identify Data Needs | 26% |
| Domain 4: Manage AI Model Development and Evaluation | 16% |
| Domain 5: Operationalize AI Solution | 17% |
Domains 2 and 3: The 52% Core
Identify Business Needs and Solutions and Identify Data Needs are jointly the largest, at 26% each. Together they account for just over half the blueprint. A candidate who is strong here but weak elsewhere is in a better position than the reverse.
- Framing a business problem so that AI is genuinely the right tool, and recognizing when it is not
- Defining measurable success criteria tied to business value
- Assessing data availability, quality, quantity, and suitability for the intended use
- Understanding data sourcing, preparation, and governance obligations before development begins
Domain 1: Support Responsible and Trustworthy AI Efforts (15%)
Smaller by weight but easy to underestimate. Responsible and trustworthy AI considerations thread through project decisions, so scenario questions in other domains often hide a responsibility angle.
- Recognizing risks such as bias, privacy exposure, and lack of transparency
- Knowing when to escalate and who needs to be involved in oversight decisions
Domain 4: Manage AI Model Development and Evaluation (16%)
This is where project managers are tested on coordinating, not coding. You need to understand model development and evaluation well enough to ask the right questions and interpret results.
- Selecting evaluation approaches that match the business objective
- Interpreting performance results and deciding whether to iterate, retrain, or revisit earlier phases
Domain 5: Operationalize AI Solution (17%)
Many AI projects stumble after the model works. This domain covers moving into real-world use and sustaining it.
- Deployment planning and integration into existing processes
- Monitoring for drift and degraded performance over time
- Planning ongoing maintenance and stakeholder adoption
For a full walkthrough of each content area, see the complete guide to all five CPMAI exam domains.
Reading Your Domain-Level Feedback
You will not receive a numerical score. Instead, PMI reports pass or fail status along with domain-level diagnostic feedback. If you pass, the feedback is a curiosity. If you do not, it becomes your most valuable planning tool.
How to use it after a result
- Identify the weakest domains first. Rank them by weight as well as by performance. A shortfall in Domain 2 or Domain 3 deserves more attention than the same shortfall in Domain 1, simply because those domains carry more of the exam.
- Revisit the underlying reasoning. If feedback points to Operationalize AI Solution, review why monitoring and maintenance decisions are made, not just the list of activities.
- Retest the weak area with fresh scenarios. A CPMAI practice test that reports performance by domain helps you confirm that a gap has closed before you pay for another attempt.
Key Takeaway
Because your official result is pass/fail with domain feedback, build your practice routine to mimic that output. Track your results per domain, not just an overall percentage, so that your preparation data resembles the report you would actually receive.
Attempts, Retakes, and the 12-Month Window
The retake rules are frequently misquoted, so the specifics matter here.
- Candidates have 12 months from purchase to obtain certification.
- Within that eligibility year, you can make up to three total exam attempts. That is three attempts in all, not three retakes on top of a first try.
- Additional attempts require payment.
- The exam outline recommends 30 days of preparation before a retake. This is advice, not a mandatory 30-day waiting period.
- If you fail three attempts within the eligibility year, the general PMI handbook requires a one-year wait from the last exam before you can reapply.
Before scheduling anything, confirm that you meet the course-completion requirement. No prior project management, technical, or AI experience or certification is required, but the 21-hour prep course must be completed first. The CPMAI requirements guide covers eligibility in detail, and the exam dates and scheduling guide helps you work backward from your deadline.
Planning Around a Criteria-Based Standard
Since there is no verified cut score to target, your preparation plan should aim for broad, durable competence, weighted toward the heaviest domains. One simple way to turn the blueprint into a schedule is to sequence your effort by weight and dependency.
Identify Business Needs and Solutions
- Largest domain at 26%; it also sets context for everything that follows
- Practice framing problems and defining success criteria
Identify Data Needs
- The other 26% domain; data feasibility questions appear in many scenarios
- Work through data quality, sourcing, and readiness cases
Model Development, Evaluation, and Operationalization
- Domains 4 and 5 together make up 33% of the blueprint
- Focus on evaluation choices, deployment, and monitoring decisions
Responsible AI Integration and Full Review
- Revisit Domain 1 across all scenarios, since responsible AI considerations cut across phases
- Take timed 120-question practice sets to rehearse the 160-minute format
This is a sample sequence, not an official schedule; adjust it to your background and the time you have. Candidates with data science experience may compress Week 2, while those new to AI may need longer there. A quick-reference summary of the essentials is available in the CPMAI cheat sheet.
Exam-Day Rules That Protect Your Result
A rule violation can invalidate a result no matter how well you know the material. Know the conditions before you sit.
Closed-book conditions
The exam is closed-book. Unauthorized materials and assistance are prohibited. The computer-based testing policy provides a built-in calculator, so you do not need to bring one.
OnVUE remote proctoring
- A system check is required before test day.
- Identity verification is part of check-in.
- You need a private testing space.
- Books, notes, paper, and physical writing materials are prohibited unless specifically approved; a digital whiteboard is provided instead.
No scheduled breaks
This exam has no scheduled breaks, which overrides generic handbook descriptions of breaks for other PMI exams. Plan your hydration, nutrition, and bathroom timing before you start the timer, and practice at least one full-length simulation so that a 160-minute sitting does not surprise you.
After You Pass: Renewal and Career Use
Earning the credential is the beginning of a maintenance cycle. Renewal requires 30 PDUs every three years, structured as follows:
- At least 18 Education PDUs
- At most 12 Giving Back PDUs
- At least four PDUs in each Talent Triangle skill area: Ways of Working, Power Skills, and Business Acumen
You must report your PDUs and pay the applicable renewal fee. Membership is optional, though fees vary by region and membership status.
On the career side, the credential signals that you can lead AI initiatives as a manager, not as a model builder. It is most relevant to project managers, product owners, business analysts, and data leaders who coordinate AI work across business and technical teams. For realistic expectations about opportunities, see our overviews of CPMAI jobs and CPMAI salary expectations, and the ROI analysis if you are weighing the investment.
Frequently Asked Questions
PMI uses a criteria-based, psychometrically established passing standard and reports pass/fail status with domain-level diagnostic feedback. No numerical passing percentage has been verified, so any specific figure you see quoted should be treated with caution.
The exam contains 120 questions in total, of which 100 are scored and 20 are unscored pretest items. You will not know which questions fall into the unscored group, so answer every question carefully.
You have 12 months from purchase and up to three total attempts within that eligibility year. Additional attempts require payment. The outline recommends 30 days of preparation before a retake, but that is a recommendation rather than a mandatory waiting period. After three failed attempts, the handbook requires a one-year wait from your last exam before reapplying.
Identify Business Needs and Solutions and Identify Data Needs are weighted 26% each, totaling 52% of the blueprint. The remaining domains are Support Responsible and Trustworthy AI Efforts (15%), Manage AI Model Development and Evaluation (16%), and Operationalize AI Solution (17%). Because the standard is criteria-based, you should still build competence in every domain.
No. This exam has no scheduled breaks, and that rule takes precedence over generic handbook language about breaks on other PMI exams. Tutorial and survey time are outside the exam timer, so the 160 minutes is testing time only.