- Does Pearson Publish a GAF Pass Rate?
- Why 700 on a 1-1000 Scale Is Not a Percentage
- What Data Actually Exists for the Exam
- What Drives Difficulty in the Four Domains
- Format Factors That Affect Outcomes
- Delivery Channel and Voucher Mistakes
- A Domain-Ordered Readiness Plan
- How to Read Third-Party Pass-Rate Claims
- Frequently Asked Questions
- Pearson and Certiport do not publish an official pass rate for the Generative AI Foundations exam.
- The passing score is 700 on a 1-1000 scaled system, not a percentage of questions answered correctly.
- The timed exam runs 50 minutes, with 40-45 questions per the tutorial and 45-50 per Mindhub's estimate.
- Four domains are covered, with no published percentage weights, so prepare evenly across all of them.
Does Pearson Publish a GAF Pass Rate?
Searching for the Generative AI Foundations pass rate usually ends in frustration, and the reason is simple: the issuer does not release one. Generative AI Foundations is a Pearson credential developed and administered through Certiport, a Pearson VUE business, as part of the Critical Career Skills (CCS) family. Certiport publishes exam objectives, a tutorial, scoring policy and voucher terms. It does not publish a first-attempt or overall pass percentage for this exam.
That matters for how you read anything you find online. Any specific figure quoted as "the GAF pass rate" is either a guess, a figure borrowed from an unrelated credential that happens to share the acronym, or a vendor's marketing number with no stated methodology. On this site we deliberately avoid quoting a number we cannot trace to the issuer. What we can do is show you what the published data does say, and how to use it to judge your own readiness. For a broader look at difficulty, see our guide on how hard the GAF exam is.
Why 700 on a 1-1000 Scale Is Not a Percentage
The one hard number tied to passing is the score threshold: 700 on Certiport's 1-1000 scaled-score system. This is the most commonly misunderstood fact about the exam, and it directly distorts how candidates think about pass rates.
A scaled score is not the same as the share of questions you answered correctly. Certiport's scoring policy states that question totals differ by exam form, and scaled scoring exists so that candidates who see different forms are held to a comparable standard. Two practical consequences follow:
- You cannot convert 700 into "70% correct" and use that as a target. The relationship between raw correct answers and the scaled result depends on the form you receive.
- A practice test that reports a simple percentage tells you something about relative strength by topic, but it cannot tell you your actual scaled score on exam day.
For a full explanation of how the threshold works, read our breakdown of the GAF passing score.
| Scoring Detail | What the Issuer Documents |
|---|---|
| Score scale | 1-1000 scaled score |
| Passing score | 700 |
| Is it a percentage correct? | No |
| Do question totals vary? | Yes, by exam form per Certiport's scoring policy |
| Published domain weights | None in the objectives outline |
What Data Actually Exists for the Exam
Since a pass rate is unavailable, the useful question becomes: what published facts can you actually use to estimate your odds? Here is the verified picture.
Format and timing
The exam is a proctored, computer-based knowledge assessment. You select answers, flag questions for review, and can revisit them before final submission. The timed portion is 50 minutes, and that excludes the tutorial and administrative seat time. Certiport's copyright-2024 tutorial specifies 40-45 questions, while the current Mindhub voucher page estimates 45-50. These sources conflict, and the scoring policy says totals differ by form, so plan for a range rather than a fixed count.
The arithmetic is worth internalizing: with roughly 40-50 questions in 50 minutes, you have a little over a minute per item. That is comfortable for a recall-style knowledge exam but tight if you stall on scenario wording. Flagging and returning is your pressure valve.
Candidate expectations
The issuer describes the exam as brand-agnostic and best suited for ages 14 and older. No bachelor's degree is required. Certiport expects candidates to have around 150 combined hours of instruction and hands-on use of generative AI tools, plus familiarity with productivity applications. Those hours describe preparation, not the exam timer or a mandatory course length. If you want the full eligibility picture, see our page on GAF requirements.
This 150-hour figure is the closest thing to an official readiness benchmark. If you have only skimmed a few AI tools, you are below the issuer's own expectation, and that gap is a more honest predictor of risk than any invented pass percentage.
What Drives Difficulty in the Four Domains
The objectives outline lists four domains and publishes no percentage weights. Without weights, you cannot safely skip a domain on the assumption that it is a small slice. Each one below is a place where candidates commonly lose points. For the deeper breakdown, see the complete guide to all four GAF content areas.
Domain 1: Generative AI Methods and Methodologies
This is the conceptual foundation. Candidates need to explain how generative systems work at a literacy level, distinguish generative approaches from other kinds of AI, and recognize the strengths and limits of different methods.
- Vocabulary precision: terms that sound interchangeable often are not, and answer choices exploit that.
- Understanding what generative models can and cannot reliably do.
- Recognizing appropriate versus inappropriate uses of generated output.
Domain 2: Basic Prompt Engineering
This domain tests whether you can construct a clear prompt. Expect questions about what makes an instruction effective: specificity, context, desired format, and role or audience framing.
- Identifying why a vague prompt produces a weak result.
- Choosing the best prompt among several similar-looking options.
- Understanding how added context and constraints change output.
Domain 3: Prompt Refinement
Where Domain 2 is about first drafts, Domain 3 is about iteration. Candidates must recognize how to diagnose a poor response and adjust the prompt to improve it.
- Reading an unsatisfying output and deciding what to change.
- Iterating toward accuracy, tone, length and structure.
- Knowing when to add examples, tighten constraints or restate the goal.
Domain 4: Ethics, Law, and Societal Impact
This domain is often underestimated by technically minded candidates. It covers responsible use, bias, privacy, intellectual property considerations and broader effects on work and society.
- Spotting bias or fairness concerns in generated content.
- Understanding privacy and data-handling implications of what you enter into a tool.
- Recognizing legal and attribution issues around AI-generated material.
The two prompting domains reward hands-on repetition, while Domains 1 and 4 reward careful reading of precise wording. Candidates strong in one style and weak in the other are the ones most likely to land near the 700 line. Our GAF cheat sheet condenses the must-know facts across all four.
Format Factors That Affect Outcomes
The scaled-score cushion and its limits
Because the result is scaled, there is no published "you may miss exactly N questions" rule you can rely on. Do not build a strategy around a target miss count. Aim for broad, consistent accuracy instead.
Proctoring rules
CATC rules require proctoring and valid identification. Candidates under 18 need parental consent. Books and personal devices must remain inaccessible during the exam, so you cannot lean on notes or a second screen. This is a closed-resource knowledge test, which means recall under time pressure matters.
Validity period
CCS certifications earned on or after June 18, 2025 are valid for five years, and recertification requires a new voucher and passing the current exam. Earlier nonexpiring certifications retain their existing status. This affects how you think about retake and renewal costs over time, covered in our GAF certification cost breakdown.
Delivery Channel and Voucher Mistakes
An overlooked cause of failed attempts is not a low score at all: it is a candidate who cannot sit the exam because they bought the wrong voucher. The published terms make this easy to get wrong.
| Voucher Type | Price | Delivery Constraint |
|---|---|---|
| Certiport single-exam voucher | USD 72 (U.S.) | Valid only through Certiport Authorized Testing Centers, in person or remotely |
| Certiport voucher with retake | USD 83 (U.S.) | Same CATC-only restriction |
| Pearson Mindhub voucher | USD 85 | Country restrictions apply; must match the country selected at checkout |
| Mindhub Pro voucher | USD 85 (web and member price) | Same country-matching requirement |
Certiport vouchers cannot be redeemed through OnVUE or a Pearson VUE Testing Center. Any CATC proctoring charge is separate. Pearson's October 1, 2024 launch announcement names an OnVUE delivery route, but the public Mindhub scheduling link currently redirects to an Adobe scheduling page, so a current GAF-specific OnVUE booking process could not be verified. The safe move is to confirm delivery and voucher compatibility with your chosen testing channel before you pay for anything. Details on scheduling live in our guide to GAF exam dates and scheduling.
A Domain-Ordered Readiness Plan
Generic schedules are less useful than ordering your preparation around how the four domains depend on each other. Use this sequence, and pair it with our GAF study guide for deeper resources.
Domain 1 first
- Build precise vocabulary; the later domains assume it.
- Practice distinguishing similar-sounding methods and terms.
Domain 2 with live tools
- Write prompts in a real generative AI tool, comparing weak and strong versions.
- Log what context and constraints changed the output.
Domain 3 iteration drills
- Take disappointing outputs and refine them across several rounds.
- Note which adjustments fix accuracy versus tone versus structure.
Domain 4, then full review
- Cover bias, privacy, intellectual property and societal impact.
- Finish with timed mixed practice at roughly one minute per question.
Domain 4 goes last in this plan because ethical judgments are easier to reason about once you have actually used the tools, but do not shortchange it. With no published weights, it can carry as much risk as any other domain.
How to Read Third-Party Pass-Rate Claims
When you encounter a claimed pass rate for this exam, test it with a few questions. Does the source name its sample size and time period? Does it state that it refers to Pearson's Generative AI Foundations specifically, rather than another credential sharing the same acronym? Is the figure traceable to Pearson or Certiport? If the answer to any of these is no, treat the number as noise.
Practice-product vendors sell preparation materials for this exam, and their reported success figures are marketing claims, not issuer statistics. They can still be useful for topic coverage, but they should not set your expectations about the real pass rate. If you are weighing whether the credential is worth the effort at all, see our ROI analysis, and for context on the roles it supports, our page on GAF jobs.
The most reliable way to gauge your own odds is to measure yourself against the published objectives and the 150-hour benchmark, then test under realistic timing. You can start with the timed drills on our main practice test site and return to it as you tighten weak domains.
Key Takeaway
Replace the question "what percentage pass?" with "am I consistently strong across all four domains at about one minute per question?" That is a measurable standard you control, and it is a far better predictor than any unverified statistic.
Frequently Asked Questions
There is no officially published pass rate. Pearson and Certiport release objectives, scoring policy and pricing, but not pass statistics for this exam. Any specific percentage you see is unverified unless it traces directly to the issuer.
You need 700 on Certiport's 1-1000 scaled-score system. This is not a percentage of correct answers, and question totals differ by exam form, so you cannot convert 700 into a fixed number of correct responses.
The timed portion is 50 minutes. Certiport's tutorial specifies 40-45 questions, while the Mindhub voucher page estimates 45-50. Totals vary by form, so prepare for a range and use the flag-and-review feature to manage time.
No. The four domains are Generative AI Methods and Methodologies, Basic Prompt Engineering, Prompt Refinement, and Ethics, Law, and Societal Impact, but the outline gives no percentage weights. Prepare evenly across all four.
No. Certiport vouchers at USD 72 or USD 83 with a retake work only through Certiport Authorized Testing Centers, and cannot be redeemed through OnVUE or a Pearson VUE Testing Center. Mindhub vouchers cost USD 85 and have country restrictions, so confirm compatibility with your testing channel before purchasing.