The ISTQB® Certified Tester – Testing with Generative AI (CT-GenAI®) is a Specialist Level certification that verifies a tester’s ability to apply Large Language Models, prompt engineering, and AI risk controls across the software testing process. The exam has 40 multiple-choice questions, runs for 60 minutes, and requires a 65% pass mark. The ISTQB® Foundation Level (CTFL®) is a mandatory prerequisite.
At a glance
| Item | Value |
|---|---|
| Certification | ISTQB® Certified Tester – Testing with Generative AI (CT-GenAI®) |
| Syllabus version | v1.1 (minor update to v1.0) |
| Released | v1.0 on 25 July 2025; v1.1 on 27 April 2026 |
| Stream | Specialist |
| Prerequisite | ISTQB® Foundation Level (CTFL®) is mandatory |
| Exam length | 60 minutes |
| Questions | 40 multiple-choice questions |
| Pass mark | 65% (26 out of 40) |
| Question style | Single-select MCQ, mapped to K1, K2, K3 |
| Time extension for non-native speakers | 15 minutes (25%) |
| Languages | English; additional languages rolled out by ISTQB® Member Boards. Verify on the board you book with. |
| Delivery | Online via iSQI FLEX (remote proctoring). Pearson VUE test centres listed as coming soon by iSQI. |
| Fee | USD 199 via ASTQB®. Other Member Boards: not published on a single indexable page at the time of writing. Verify on istqb.org before booking. |
| Validity | Lifetime. The certification does not expire. |
What is CT-GenAI®?
CT-GenAI® is a Specialist Level credential offered by ISTQB® that certifies a tester’s ability to apply Generative AI, Large Language Models, and AI-powered tooling across the software testing process. It sits in the Specialist branch of the ISTQB® portfolio, alongside other applied-domain certifications such as CT-AI® (testing AI-based systems) and CT-MAT® (mobile testing).
The syllabus was formally approved by the ISTQB® General Assembly on 25 July 2025 and announced publicly in late July 2025. A minor update, v1.1, was released on 27 April 2026 with terminology refinements (for example, “few-shot” updated to “one-shot”), clarifications to evaluation metrics, and added context for LLM-powered agents. The v1.1 update does not change the structure, learning objectives, or examinable scope, and accredited training providers were not required to reaccredit.
CT-GenAI® is not a replacement for CT-AI®. The two cover different problem spaces. CT-AI® focuses on how to test AI-based systems (input data testing, ML model testing, performance metrics, fairness). CT-GenAI® focuses on how to use Generative AI to support and transform the testing process itself. ISTQB® explicitly recommends CT-GenAI® for candidates interested in applying GenAI to testing activities, and CT-AI® for those who test AI systems.
Who should take CT-GenAI®?
This certification is built for working testing professionals who want to apply GenAI tools in their day-to-day work, including:
- Manual and exploratory testers who use ChatGPT, Copilot, Claude, or Gemini for test ideas, test data, or acceptance criteria
- Test automation engineers and SDETs building or maintaining LLM-assisted automation
- Test analysts working with user stories, acceptance criteria, and Gherkin scenarios
- Test managers and QA leads defining a GenAI adoption roadmap for their teams
- UAT testers and business analysts who need to question AI-generated test artefacts
- Developers and DevTest engineers who write or maintain LLM-powered test tooling
It is also a useful baseline for project managers, quality managers, and IT directors who need to understand the risks and limits of GenAI before approving it for use in regulated test environments.
Who should not take it. If you have not yet passed the ISTQB® Foundation Level, CT-GenAI® is not your next step, since CTFL® is a mandatory prerequisite. If your goal is to test AI/ML systems (model drift, training-data validation, fairness metrics), CT-AI® v2.0 is the correct path, not CT-GenAI®. And if you have no exposure to LLMs in any form, expect a steep learning curve in Chapter 2 (prompt engineering, 365 minutes of instruction time).
Prerequisites and eligibility
- Mandatory prerequisite. ISTQB® Foundation Level (CTFL®) certificate. This is stated explicitly in the CT-GenAI® syllabus, section 0.6.
- Recommended experience. Practical use of at least one mainstream LLM tool (ChatGPT, Claude, Copilot, Gemini) for a testing task. The syllabus assumes you can run a prompt and read the output critically.
- Recommended prior reading. The ISTQB® Glossary and the CTFL® v4.0 syllabus, both available free on istqb.org. The syllabus uses CTFL® terminology throughout.
- No experience or age requirements are stated by ISTQB® beyond the CTFL® prerequisite.
Exam structure and rules
Format and length
The CT-GenAI® exam is a 60-minute, closed-book test of 40 multiple-choice questions. There is no oral or practical component. The hands-on objectives in the syllabus are training requirements, not exam requirements.
One-line summary: CT-GenAI® is a 60-minute, 40-question multiple-choice exam.
Question types and K-level distribution
Questions are mapped to three cognitive levels defined in the syllabus: K1 (Remember), K2 (Understand), and K3 (Apply). K3 questions are scenario-based and typically ask you to choose the correct prompt engineering technique, mitigation strategy, or evaluation metric for a given test situation. The syllabus introduction, hands-on objectives, and appendices are explicitly not examinable.
One-line summary: CT-GenAI® questions cover K1, K2, and K3, with K3 scenarios drawn from real testing tasks.
Passing score
You need 65% to pass, which equates to 26 correct answers out of 40. There is no negative marking on standard ISTQB® exams. Confirm the exact threshold with the Member Board you book through, since the official “Exam Structures and Rules” document referenced in the syllabus is the authoritative source.
One-line summary: The CT-GenAI® passing score is 65%, or 26 out of 40.
Time extension policy
Candidates taking the exam in a language that is not their native language are entitled to a 25% time extension, which equals 15 extra minutes (total 75 minutes). This applies to most non-native English speakers sitting the English-language exam.
One-line summary: Non-native speakers get 15 additional minutes, for a total of 75 minutes.
Languages available
CT-GenAI® launched in English. iSQI lists Latin American Spanish and European Spanish as available exam versions. Other languages are rolled out gradually by individual ISTQB® Member Boards. Not stated on a single official summary page at the time of writing for the complete language list. Verify on the board you book with before paying.
One-line summary: English is confirmed; iSQI lists two Spanish variants. Other languages are board-dependent.
Delivery options
iSQI delivers CT-GenAI® online through its iSQI FLEX remote-proctored platform. iSQI’s product page lists Pearson VUE test centre delivery as “coming soon” at the time of writing. ASTQB® delivers the exam via Kryterion centres and remote proctoring in the United States. Other Member Boards use their own logistics, so check before assuming a specific delivery method.
One-line summary: Online remote proctoring is the primary delivery method today, with in-person centres expanding.
Retake policy
Retake rules are set by the Member Board that issued your voucher, not by ISTQB® centrally. Most boards allow a paid retake without a mandatory waiting period. Verify on your board’s terms and conditions before booking.
One-line summary: Retakes are paid and board-specific. Most boards allow a quick retake.
Cost
ASTQB® publishes a fixed CT-GenAI® price of USD 199 on astqb.org. Fees from other Member Boards (BCS in the UK, ITB in India, iSQI globally, GASQ in Germany, ANZTB in Australia and New Zealand) are not published on a single indexable page at the time of writing. They typically fall in the USD 150 to USD 250 range for Specialist exams. Verify on istqb.org’s Member Board directory before booking.
One-line summary: USD 199 in the United States via ASTQB®; verify locally for other regions.
Certificate validity and renewal
The CT-GenAI® certificate has lifetime validity. ISTQB® does not require renewal, continuing education credits, or expiry-based recertification at the Specialist Level. Only Expert Level certifications expire and require recertification.
One-line summary: CT-GenAI® is valid for life and does not require renewal.
Syllabus breakdown, chapter by chapter
The CT-GenAI® v1.1 syllabus has five examinable chapters and requires a minimum of 13.6 hours (815 minutes) of accredited instruction time. The official chapter titles, time budgets, and learning objectives below are taken directly from the ISTQB® syllabus.
Chapter 1: Introduction to Generative AI for Software Testing
Suggested study time: 100 minutes.
This chapter sets the conceptual foundation. It covers the AI spectrum (symbolic AI, classical machine learning, deep learning, GenAI), the basics of LLMs and Small Language Models, and the distinctions between foundation, instruction-tuned, and reasoning models. It also introduces multimodal LLMs and vision-language models.
Why it matters in real testing work: a tester who cannot distinguish a reasoning LLM from an instruction-tuned one will pick the wrong tool for the wrong job. Reasoning models are slower and more expensive, but they outperform on test-planning and prioritisation tasks. Instruction-tuned models are cheaper for high-volume test data generation. The chapter also frames concepts like tokenisation and context windows, which directly affect how much test log data you can fit into a single prompt.
K-levels involved: mostly K1 and K2. Expect recall-style questions on definitions (tokenisation, embeddings, transformer, context window) and explanation-style questions on the differences between LLM types.
Chapter 2: Prompt Engineering for Effective Software Testing
Suggested study time: 365 minutes. This is the heaviest chapter, accounting for 45% of training time.
This chapter is the operational core of the certification. It defines the six-component structured prompt (role, context, instruction, input data, constraints, output format), three core prompting techniques (prompt chaining, few-shot prompting, meta prompting), and the distinction between system prompts and user prompts. It then walks through applying these techniques to test analysis, test design, test implementation, automated regression testing, and test monitoring. The chapter closes with evaluation metrics for GenAI output (accuracy, precision, recall, relevance, diversity, execution success rate, time efficiency) and techniques for iteratively refining prompts.
Why it matters in real testing work: every other chapter assumes you can write a structured prompt. A team using “make me some test cases for this user story” will get inconsistent, hallucinating output. A team using a six-component prompt with a Gherkin example and a constraint on output format will get consistent, reviewable artefacts. The exam tests this distinction directly.
K-levels involved: K2 for the prompt structure and the techniques, K3 for applying them to specific test tasks. Expect scenario questions where you must select the best prompting technique for a given situation (for example, repetitive Gherkin generation calls for few-shot, complex multi-step analysis calls for prompt chaining).
Chapter 3: Managing Risks of Generative AI in Software Testing
Suggested study time: 160 minutes.
This chapter covers four risk areas: hallucinations and reasoning errors and biases; data privacy and security; energy consumption and environmental impact; and AI regulations and standards. It introduces the temperature parameter and random seeds as mitigation techniques for non-deterministic LLM behaviour. It also names specific frameworks that an exam candidate is expected to recall, including ISO/IEC 42001:2023, ISO/IEC 23053:2022, the EU AI Act, and the NIST AI Risk Management Framework.
Why it matters in real testing work: this is where regulated industries earn or lose the right to use GenAI. A bank, a hospital, or a defence contractor cannot send production-equivalent test data to a public LLM without violating GDPR, HIPAA, or sector-specific rules. The chapter teaches data minimisation, anonymisation, secure-environment selection, and human review as a layered defence.
K-levels involved: K1 for definitions of hallucination, reasoning error, and bias; K2 for explaining risks and mitigations; K3 for identifying hallucinations in LLM output. Expect at least one question on the EU AI Act and at least one on distinguishing hallucinations from reasoning errors.
Chapter 4: LLM-Powered Test Infrastructure for Software Testing
Suggested study time: 110 minutes.
This chapter covers the architectural side: how LLM-powered test tools are built, how Retrieval-Augmented Generation (RAG) works, the role of LLM-powered agents in automating test processes, fine-tuning LLMs for testing tasks, and LLMOps. It explains vector databases, embeddings, semantic similarity retrieval, and the difference between autonomous and semi-autonomous agents.
Why it matters in real testing work: most enterprise GenAI testing tools in 2026 are RAG-based, because they connect a general-purpose LLM to a company’s private test cases, defect history, and requirements. Understanding what RAG can and cannot do prevents teams from over-trusting an agent that has no access to the real specification.
K-levels involved: mainly K2. Expect questions on the components of a RAG pipeline (chunking, embedding, retrieval, generation), on what fine-tuning is for, and on the role of LLMOps in deploying and monitoring LLMs in test environments.
Chapter 5: Deploying and Integrating Generative AI in Test Organizations
Suggested study time: 80 minutes.
The lightest chapter by time, but examinable. It covers the risks of Shadow AI, the structure of a GenAI adoption strategy, criteria for selecting LLMs and SLMs for test tasks, and the phased approach to rolling out GenAI in a test organisation. It also covers change management, essential skills for testers in GenAI-enabled teams, and how test processes evolve.
Why it matters in real testing work: most CT-GenAI® candidates who fail in the workplace, not the exam, fail because they treat GenAI adoption as a tooling decision rather than a process change. This chapter teaches the test manager’s view.
K-levels involved: K1 and K2. Expect questions on Shadow AI risks, on the phases of GenAI adoption, and on the skills test teams need to build.
How to prepare for CT-GenAI®
Recommended study plan
4-week plan (intensive, for candidates already comfortable with LLMs). Week 1: Chapters 1 and 5 (foundation and organisational view). Week 2: Chapter 2, part 1 (prompt structure and techniques). Week 3: Chapter 2, part 2 (applying techniques to specific test tasks) plus Chapter 4. Week 4: Chapter 3, full syllabus review, two timed mock exams.
8-week plan (balanced, recommended for most candidates). Two weeks per major chapter, with the heaviest weight on Chapter 2. End the eighth week with three full timed mocks and a debrief.
12-week plan (for candidates new to LLMs). Spend the first three weeks getting hands-on with a public LLM before opening the syllabus. Write at least 30 prompts for real test artefacts (acceptance criteria, test cases, test data, regression test review). Then follow the 8-week plan for the remaining nine weeks.
Official materials
- The official CT-GenAI® syllabus (v1.0 PDF as the verifiable artefact; replace with the v1.1 link once published on istqb.org)
- The ISTQB® Glossary (free download, defines every term the exam can use)
- The official CT-GenAI® sample exam (available in the Download section of the iSQI CT-GenAI® product page and through other Member Boards)
Self-study vs accredited training
Accredited training requires a minimum of 13.6 hours of instruction by an ISTQB®-accredited provider, and most courses run for three full days. The advantage is hands-on labs and instructor feedback on prompts. The disadvantage is cost: training plus exam often totals USD 1,500 or more.
Self-study is allowed and increasingly popular for Specialist Level exams. The syllabus is a specification, not a textbook, so self-study candidates need a supplementary guide. Self-study costs are limited to the exam fee plus any practice materials you buy.
Honest tradeoff: self-study works well if you already use LLMs daily. Accredited training is the safer choice if you have read about prompt engineering but never built a multi-step prompt chain in anger.
Practice exams
Use the official sample exam from ISTQB® first. After that, multiple board-affiliated and third-party providers publish CT-GenAI® practice tests. Verify any provider’s claim by checking their accreditation status on the relevant Member Board’s site. Do not use braindump sites; ISTQB® content is copyrighted, and leaked-question sites do not reliably reflect the current syllabus version.
How to read the sample exam debrief
The official sample exam comes with an answer key that explains why each option is correct and why the others are not. Read the explanation even for questions you got right; the reasoning often reveals a distinction (for example, hallucination vs reasoning error) that the exam tests in several variants.
Mistakes that cause people to fail
- Confusing CT-AI® with CT-GenAI®. Candidates study the wrong syllabus. CT-AI® is about testing AI systems. CT-GenAI® is about using AI to test.
- Skipping Chapter 2 sub-techniques. Many candidates can define few-shot prompting but cannot pick it over meta prompting in a scenario question. The exam tests selection, not definition.
- Ignoring the regulations table. ISO/IEC 42001:2023, ISO/IEC 23053:2022, the EU AI Act, and NIST AI RMF are all explicitly named in the syllabus and are fair game for K1 questions.
- Underestimating the temperature and seed concepts. These appear in Chapter 3 as mitigations for non-deterministic behaviour and are frequently tested.
- Treating the hands-on objectives as examinable. They are not. They are training requirements, and the introduction and appendices are also out of scope.
How CT-GenAI® compares with related certifications
| Feature | CT-GenAI® | CT-AI® v2.0 | CTFL® v4.0 |
|---|---|---|---|
| Stream | Specialist | Specialist | Core (Foundation) |
| Prerequisite | CTFL® | CTFL® | None |
| Exam length | 60 minutes | 60 minutes | 60 minutes |
| Questions | 40 | 40 | 40 |
| Pass mark | 65% | 65% | 65% |
| Audience | Testers using GenAI for testing | Testers testing AI/ML systems | Anyone entering testing |
| What it certifies | Applying LLMs, prompt engineering, AI risk controls to the test process | Testing ML models, input data, fairness, ML lifecycle | Core testing fundamentals across the lifecycle |
| Indicative fee (US, via ASTQB®) | USD 199 | USD 199 | USD 229 |
The most frequent question prospective candidates ask is whether they need both CT-AI® and CT-GenAI®. The honest answer: only if your role spans both sides. A QA lead in a company building an AI product genuinely needs CT-AI®. A QA lead in a company using AI tools to improve its testing genuinely needs CT-GenAI®. Most testers in 2026 are in the second group.
Career impact and recognition
CT-GenAI® is one of the most recent Specialist Level certifications, so longitudinal salary data is not yet meaningful. What is verifiable: ISTQB® certifications appear in QA job descriptions on LinkedIn and Indeed for roles such as Test Automation Engineer, SDET, AI Test Engineer, and QA Lead. Searches for “CT-GenAI” and “ISTQB Generative AI” on those platforms return live results across the US, UK, India, Germany, and the Netherlands as of mid-2026.
ISTQB® does not publish pass rates for any of its certifications, and no third party reliably tracks them either. Treat any specific pass-rate figure with scepticism unless it cites an audited source.
Two practical signals matter more than speculative salary numbers. First, accredited training providers globally added CT-GenAI® to their catalogues within months of the July 2025 release, which suggests employer demand. Second, the v1.1 update in April 2026 was a minor refinement rather than a structural rewrite, which suggests the certification has stabilised and is unlikely to be replaced soon.
Official downloads and resources
- ISTQB® CT-GenAI® official certification page on istqb.org
- ISTQB® CT-GenAI® syllabus v1.0 PDF (v1.1 replaces this when published on istqb.org)
- ISTQB® Glossary (free, defines all examinable terminology)
- ISTQB® Member Board directory (find the board that issues your voucher)
- iSQI CT-GenAI® product page (booking, sample exam, current pricing)
Frequently asked questions
Is CT-GenAI® worth it in 2026?
Yes, for testers who already use or plan to use GenAI tools in their daily work. CT-GenAI® is the only globally recognised certification specifically for applying LLMs and prompt engineering to the testing process. It also signals to employers that you understand the data privacy, hallucination, and regulatory risks that come with GenAI in regulated industries.
How hard is the CT-GenAI® exam?
Moderately hard. The exam is shorter (40 questions, 60 minutes) than CTFL® has been historically, but Chapter 2 carries 45% of the training weight, and the K3 scenario questions reward genuine hands-on prompt engineering experience. Candidates who have used LLMs only as chatbots usually find the exam harder than expected.
How much does the CT-GenAI® exam cost?
ASTQB® lists the CT-GenAI® exam at USD 199 in the United States. Other Member Boards publish their own fees, which generally fall between USD 150 and USD 250. Not published on a single official summary page at the time of writing. Verify on istqb.org’s Member Board directory before booking.
How long does it take to prepare for CT-GenAI®?
Most candidates need four to twelve weeks. Four weeks suits testers already comfortable with LLMs and prompt engineering. Eight weeks is a balanced default. Twelve weeks is realistic for candidates new to GenAI tools who need time to build hands-on familiarity before opening the syllabus.
Does CT-GenAI® expire?
No. CT-GenAI® has lifetime validity. ISTQB® does not require renewal, continuing education credits, or expiry-based recertification for Specialist Level certifications. Only Expert Level certifications expire.
Can I take CT-GenAI® online?
Yes. iSQI delivers CT-GenAI® online through its iSQI FLEX remote-proctored platform, and ASTQB® offers remote-proctored delivery in the United States. iSQI also lists Pearson VUE test centre delivery as coming soon. Other Member Boards offer board-specific online or in-person options.
What is the pass rate for CT-GenAI®?
ISTQB® does not publish pass rates for any of its certifications, and no audited third party publishes reliable CT-GenAI® pass-rate data. Treat any specific percentage you see online with scepticism.
What is the difference between CT-GenAI® and CT-AI®?
CT-GenAI® is about using Generative AI to support and transform the testing process. CT-AI® is about testing AI-based systems, including ML model testing and input data testing. ISTQB® explicitly recommends CT-GenAI® for candidates focused on applying GenAI to testing tasks, and CT-AI® v2.0 for candidates testing AI products.
Do employers recognise CT-GenAI®?
Yes. ISTQB® is the de facto global standard for software testing certifications and is recognised in over 130 countries. CT-GenAI® appears in job postings on LinkedIn and Indeed for roles such as QA Lead, Test Automation Engineer, SDET, and AI Test Engineer in markets including the US, UK, India, Germany, and Australia.
Which book is best for CT-GenAI®?
Not stated as a single official recommendation at the time of writing. ISTQB® has not endorsed a specific book for CT-GenAI®. The verifiable primary sources remain the official syllabus, the ISTQB® Glossary, and the official sample exam from your Member Board. Verify any third-party book against the v1.1 syllabus before buying.
Do I need to pass CTFL® first?
Yes. The ISTQB® Foundation Level (CTFL®) certificate is a mandatory prerequisite stated in section 0.6 of the CT-GenAI® syllabus. You cannot book CT-GenAI® without an active CTFL® certificate.
Key takeaways
- CT-GenAI® is the ISTQB® Specialist Level certification for applying Generative AI to the software testing process.
- The exam has 40 multiple-choice questions, runs for 60 minutes, and requires 65% to pass.
- The ISTQB® Foundation Level (CTFL®) is a mandatory prerequisite, confirmed in syllabus section 0.6.
- The v1.1 syllabus was released on 27 April 2026 as a minor update to v1.0 and is the current examinable version.
- Chapter 2, Prompt Engineering, carries 365 of the 815 examinable minutes and is the heaviest single topic.
- The certificate is valid for life and does not require renewal.
- CT-GenAI® and CT-AI® are complementary, not interchangeable, certifications.
Next steps
To book the exam, start at the official CT-GenAI® page on istqb.org and follow the link to the Member Board that issues vouchers in your country. United States candidates can book via ASTQB®. UK candidates can book via BCS. Global candidates without a national board can book via iSQI.
Before paying, download the syllabus and run through at least one full official sample exam.
Last reviewed on 2026-05-21 against the ISTQB® CT-GenAI® v1.0 syllabus PDF and the ISTQB® press release for v1.1.
istqb.com is an independent educational resource. ISTQB®, CTFL®, CT-AI®, and CT-GenAI® are registered trademarks of the International Software Testing Qualifications Board. This article summarises publicly available information and is not an official ISTQB® publication. Always verify exam details on istqb.org before booking.