CT-AI v2.0 or CT-GenAI v1.1: Which ISTQB Certification Is Right for You?

Two ISTQB certifications carry AI in the name. Both sit at the Specialist level. Both require the same Foundation Level prerequisite. Both talk about large language models, prompt engineering, and Retrieval Augmented Generation. No wonder testers keep asking which one to take.

Here is the line that settles it. The Certified Tester AI Testing (CT-AI) v2.0 certification teaches you to test AI systems. The Certified Tester Testing with Generative AI (CT-GenAI) v1.1 certification teaches you to test with AI. One treats the AI as the product under test. The other treats AI as a tool in your hand.

That single difference decides the right choice for you. Pick based on what your job actually is, not on which title sounds more current. This guide shows you exactly which profile each exam fits, gives you a five-question decision framework that ends with a clear answer, and lays out the prep reality for each.

The One Distinction That Ends the Confusion

Most of the confusion comes from a recent change. The original CT-AI v1.0 syllabus from 2021 included a chapter on using AI to support testing. ISTQB removed that content from CT-AI v2.0 and moved that whole subject into CT-GenAI. So the two certifications now split cleanly along one axis: testing AI, versus testing with AI.

What CT-AI v2.0 Actually Certifies

CT-AI v2.0 is for the tester who is responsible for the quality of an AI-based product. ISTQB formally released the syllabus through its General Assembly on April 17, 2026, and member boards began offering the exam from around April 21. It replaces v1.0, which retires in April 2027.

The v2.0 syllabus was restructured around the lifecycle of a machine learning system. It covers input data quality testing, model-level testing, and system-level testing, and it introduces techniques like metamorphic testing and red teaming that apply directly to validating machine learning and generative AI products. You learn to assess training data, detect bias, evaluate model performance against the right metrics, manage drift once the model is in production, and probe generative outputs for failure. ISTQB describes the new version as having a stronger focus on how AI systems are tested “in practice.”

If your deliverable is a statement like “this model behaves correctly and safely,” CT-AI is your certification. For the full breakdown, see our CT-AI v2.0 certification guide and the detailed v2.0 versus v1.0 changes.

What CT-GenAI v1.1 Actually Certifies

CT-GenAI is for any tester who wants to use generative AI to do testing work better and faster, on any system. The system under test can be a conventional, fully deterministic web or mobile application. The certification is about the tools you bring to the job, not the nature of the product.

ISTQB approved the v1.0 syllabus in July 2025 and released a minor v1.1 update on April 27, 2026. The update was small. It refined terminology, for example moving from “few-shot” to “one-shot” in places, clarified evaluation metrics, and added context for LLM-powered agents and AI-assisted testing. It did not change the structure, the learning objectives, or the examinable scope, and accredited training providers were not required to reaccredit.

The content covers LLM fundamentals, prompt engineering, RAG pipelines and agents, evaluating AI-generated content for factuality and bias, managing risks like hallucinations and data privacy, and adopting generative AI across a testing organization. Read more on our CT-GenAI v1.1 certification page.

Side by Side: CT-AI v2.0 vs CT-GenAI v1.1

Dimension	CT-AI v2.0	CT-GenAI v1.1
Core question it answers	How do I test an AI system?	How do I test using AI tools?
Role of AI	The system under test	The tool you test with
What it certifies	Validating machine learning and generative AI products	Applying generative AI across testing work
Best for	Testers of AI and ML products	Testers wanting AI-assisted workflows
Typical job titles	AI Tester, ML Quality Engineer, AI QA	Any QA, SDET, Test Analyst, or Test Lead using AI
Level and prerequisite	Specialist, requires CTFL	Specialist, requires CTFL
Exam	40 questions, 44 points, 29 to pass (65 percent), 60 minutes	40 questions, 46 points, 30 to pass (65 percent), 60 minutes
Minimum training time	19.5 hours (three days)	About 13.6 hours
Released	General Assembly approval April 17, 2026	v1.1 released April 27, 2026
Validity	Lifetime	Lifetime
How fast the content dates	Moderate, built on durable testing principles plus ML	High, tied to fast-moving tools

Where the Two Genuinely Overlap, and Where They Do Not

Both touch large language models, RAG, and generative AI risks, which is the surface that fools people. The direction is opposite. CT-AI looks at a RAG chatbot and asks how to validate that it answers correctly and safely. CT-GenAI looks at the same RAG concept and asks how a tester can use it to support testing.

The shared ground is thin. By our analysis of the two syllabi, fewer than ten percent of the learning objectives overlap. Studying for one does not prepare you for the other. Several so-called CT-GenAI study guides we have reviewed were actually relabeled CT-AI material, which is exactly why candidates walk into the wrong exam underprepared.

Which One Fits Your Role

This is the part that matters. Match the certification to the work you are paid to do.

Choose CT-AI v2.0 If

You test products that contain a model. That includes recommendation engines, fraud and anomaly detection, computer vision, demand forecasting, autonomous or driver-assistance systems, and customer-facing LLM or RAG features. Your job is to find where the model is wrong, biased, brittle, or drifting.

You work in a regulated domain that deploys AI, such as healthcare, finance, automotive, or insurance, where you must demonstrate that a model meets quality and fairness requirements before it ships.

You sit close to data and ML engineering, or you want to. The career titles this maps to are AI Tester, ML Quality Engineer, and AI QA Lead. These roles are growing as organizations move models into production and discover that conventional test design does not catch model failures.

A concrete example. A tester at a bank validating the fraud detection model, checking its precision and recall, its behavior on edge cases, and its fairness across customer segments, needs CT-AI v2.0.

Choose CT-GenAI v1.1 If

You test conventional software and you want generative AI to make you faster and more thorough. You want to use LLMs to draft test cases, generate test data, summarize requirements into checks, assist with automation scripts, and accelerate exploratory and regression work, all while managing the risks those tools introduce.

You are any practicing tester, on any stack, who wants to stay employable as AI reshapes the workflow. CT-GenAI does not assume you test AI products at all. It applies to a web tester, a mobile tester, an API tester, an automation engineer, or a test lead.

A concrete example. A tester at the same bank, testing the mobile banking app and using an LLM to draft and maintain test cases and spot gaps in coverage, needs CT-GenAI v1.1. Same employer, same person possibly, different deliverable, different certification.

A Five-Question Decision Framework

Run these in order. Stop at the first clear answer.

1. Is the thing you are paid to test an AI or ML system, or a feature whose model behavior you must validate? If yes, choose CT-AI v2.0.
2. Do you mainly test conventional software and want to use AI to do that work faster and better? If yes, choose CT-GenAI v1.1.
3. Does your employer or a regulated environment require a specific certification? If so, that requirement wins over everything else.
4. Are both true, meaning you test an AI product and you also want an AI-assisted workflow? Map to your primary responsibility. If success means “the AI works correctly,” take CT-AI. If success means “the testing got done faster and better,” take CT-GenAI.
5. Still unsure, and you simply want to stay current and employable? Choose CT-GenAI v1.1. It applies to every tester regardless of what they test, so it has the widest reach. CT-AI pays off most when you actually test AI systems.

What to Do If Both Apply

You can hold both. They share the CTFL prerequisite, neither expires, and they do not conflict. The only real question is order. For most testers, CT-GenAI is the faster first step because it delivers daily productivity gains and has a lower barrier, then CT-AI follows once you move onto AI products. We cover the ordering question in full in CT-AI vs CT-GenAI: which to take first.

Effort, Prerequisites, and Prep Reality

Both certifications require the ISTQB Foundation Level (CTFL) first. If you do not hold it yet, start there with our CTFL v4.0 study guide. The exam format is almost identical between the two. Each is 40 questions in 60 minutes, with 25 percent extra time in a non-native language, and a 65 percent pass mark. CT-GenAI draws on the K1 to K3 cognitive levels. CT-AI v2.0 sits mostly at K2 with only a few K3 application questions and no standalone recall questions, so it leans on understanding and applying concepts rather than memorizing terms. The real difference is not the exam, it is the preparation.

Time and Difficulty

CT-AI v2.0 is the heavier lift. The minimum training time is 19.5 hours across three days, and the material leans into statistical and data concepts, model evaluation metrics, and techniques like metamorphic testing. ISTQB recommends around six months of practical experience in testing, development, or data science before you sit it. If you have no exposure to machine learning, budget extra time.

CT-GenAI v1.1 is lighter on paper, with about 13.6 hours of minimum instruction and a more conceptual plus hands-on prompting focus. The catch is volatility. The content is tied to tools that change quickly, so the knowledge goes stale if you do not apply it. Treat it as a skill you keep using, not a certificate you frame and forget.

How to Prepare for Each

For CT-GenAI, our CT-GenAI v1.1 study guide is live and aligned to the current syllabus, and the launch write-up explains how the material maps to the exam. Build a small reference set of questions and expected answers in a domain you know, then practice prompts and adversarial variants against a real model so you understand how behavior shifts.

For CT-AI, the dedicated v2.0 study package is launching soon. Until then, use the CT-AI v2.0 certification guide for the exam overview, and review the official syllabus directly. If your background is light on data and ML, pair your reading with hands-on exposure to a real model and its evaluation metrics.

For the wider context on how these certifications fit the job market, our analysis of whether ISTQB still matters in the age of AI is worth a read. The headline pattern from 2026 industry coverage is a wide gap between intent and execution, with a large majority of organizations naming AI testing a priority while only a small fraction have implemented it. The tester who understands both structured testing and AI is the one who closes that gap.

You can browse every active certification and its preparation package on our study materials hub.

Frequently Asked Questions

Is CT-GenAI a newer version of CT-AI? No. They are separate certifications with different goals. CT-AI certifies testing AI systems. CT-GenAI certifies using generative AI to support testing. They share fewer than ten percent of their learning objectives.

Do I need CTFL for both? Yes. The ISTQB Foundation Level certificate is a mandatory prerequisite for both exams.

Can I take both certifications? Yes. Both are lifetime credentials with no conflict. The practical question is sequencing, which we cover in our which-to-take-first guide.

Does CT-AI v2.0 now cover generative AI? Yes, but from the testing-the-product side. CT-AI v2.0 includes how to test generative AI systems, such as red teaming their outputs. It no longer covers using generative AI as a testing tool, which now lives in CT-GenAI.

Which exam is easier? For a general tester, CT-GenAI usually requires less preparation because the concepts are more accessible. CT-AI is harder if you have no machine learning or data background, since it expects comfort with model evaluation and statistical thinking.

Is CT-AI v1.0 still worth taking? No. The v1.0 syllabus retires in April 2027 and has been replaced. New candidates should target CT-AI v2.0 and study material aligned to it.