AWSBeginnerUpdated October 2026

AIF-C01 AWS Certified AI Practitioner: 5 Practice Tests

325 questions on Amazon Bedrock, RAG, prompt engineering, responsible AI and AI security. Every option explained.

325questions
5practice exams
Available inEnglish

$ 14,99 incl. VAT where applicable

Instant access after payment.

Try 5 free sample questions

Can you pass the AWS Certified AI Practitioner exam right now? That is the only question these tests answer, and they answer it five times over.

This practice set is 325 questions across five timed practice exams. Each test is 65 questions in 90 minutes or less, the exact shape of the real AIF-C01 exam, and each follows the official domain weighting precisely: 13 questions on Fundamentals of AI and ML, 16 on Fundamentals of GenAI, 18 on Applications of Foundation Models, 9 on Guidelines for Responsible AI, and 9 on Security, Compliance and Governance. Every question is written against the AIF-C01 exam guide published on 30 April 2026.

That date matters more than it sounds. AIF-C01 has no retired predecessor, but the exam was substantially rewritten in 2026. Agentic AI, the Model Context Protocol, context engineering, token-based pricing, prompt versioning and hallucination grounding all entered the syllabus, while several services that older practice material still tests were removed from scope. Question banks last updated in 2024 or 2025 cannot contain the new material and still contain the old. Preparing on them means preparing for an exam that no longer exists.

The five tests rise in difficulty. Test 1 covers the vocabulary and the core distinctions. Test 2 puts you in short practical situations where the wrong answer is the neighbouring service. Test 3 gives you two defensible options and one constraint that decides between them. Test 4 goes into parameters, pipelines and controls. Test 5 is a full-length simulation at real exam timing, with the longest scenarios and the domains mixed.

Every answer option has its own explanation saying why it is right or wrong, and telling you what a wrong option actually is. Every question closes with a link to the AWS documentation page the answer rests on. After each attempt PrepKiwi reports your score by domain, so you know where to go next. The pass mark is set at 80 percent for tests 1 to 4 and 85 percent for test 5, well above what AWS requires, so clearing them means something.

These are practice exams, not a course. They test and explain; they do not teach the AIF-C01 syllabus from the ground up, and they contain no real exam questions. Use them alongside the free AWS material, and they will tell you exactly when you are ready.

Groundwork: AI, ML, and GenAI Foundations

65 questions · 110 minutes · pass mark 80%

Start here. This test covers the vocabulary and the core distinctions the AIF-C01 exam assumes you already have: artificial intelligence against machine learning against deep learning against generative AI, supervised against unsupervised against reinforcement learning, training against inference, and the AWS AI services matched to the tasks they were built for. It also covers the generative AI building blocks - tokens, embeddings, prompts, foundation models, retrieval augmented generation - and the responsible AI and security terms that carry 28 percent of the scored content between them. Questions are short and direct: one concept each, no layered scenarios. You get 110 minutes for 65 questions, which is more room than the real exam allows, because at this stage the point is to find the gaps rather than to work under pressure. The pass mark is 80 percent, higher than the real exam, so that a pass here means you have genuinely cleared the foundations rather than scraped through them. Every answer option carries its own explanation telling you why it is right or wrong, and every question ends with a link to the AWS documentation page the answer comes from. Take this one before you study, not after: a cold score tells you which of the five exam domains needs your attention.

In Practice: Choosing the Right AWS AI Service

65 questions · 105 minutes · pass mark 80%

The AIF-C01 exam rarely asks what a service is. It asks which one fits the situation in front of you, and the wrong answers are almost always the neighbouring service that does something similar. This test is built around exactly those confusions: Amazon Comprehend against Amazon Textract, Amazon Polly against Amazon Transcribe, Amazon Lex against both, batch against real-time against asynchronous against serverless inference, retrieval augmented generation against fine-tuning, and few-shot against zero-shot prompting. Each question gives you a short, concrete situation and asks for the service, technique or setting that meets it. You get 105 minutes for 65 questions and need 80 percent to pass. Where an answer turns on a distinction that is easy to lose, the explanation for the wrong option tells you what that option actually does, so the difference sticks. Every question links to the AWS documentation page behind the answer. Work through this test after you have covered the material once. If you find yourself picking the right family of service but the wrong member of it, that is the specific gap this test is designed to surface.

Trade-offs: Cost, Latency, and Customization

65 questions · 100 minutes · pass mark 80%

At this level two answers are defensible and one stated constraint decides between them. A data residency rule, a latency budget, a cost ceiling, how often the underlying data changes, how much labelled data you have, an explainability regulation, a database you already run - each question names the constraint that settles it, and the wrong answers are options that would have been right under a different constraint. This is where retrieval augmented generation against fine-tuning against continued pre-training against distillation gets decided properly, where precision against recall stops being a definition, and where the cost and availability trade-offs of Amazon Bedrock become choices rather than facts. Five questions here simulate the ordering and matching question types that AWS uses on the real exam and that most practice tests skip entirely. You get 100 minutes for 65 questions and need 80 percent. Explanations for the near-miss options say explicitly what condition would have made them correct, which is the part worth reading even when you answered correctly. Every question links to its source in the AWS documentation.

Under the Hood: Parameters, Pipelines, and Controls

65 questions · 95 minutes · pass mark 80%

This is the most technical of the five. It asks what a setting actually does, what order the steps go in, and how a mechanism works: temperature against top-p against top-k, stop sequences and maximum output length, how a knowledge base ingests and indexes documents, what separates a retrieval call from a retrieve-and-generate call, how prompt versioning works, what an automatic evaluation job measures against what a human evaluation job measures, how Provisioned Throughput is bought and what it buys, and what distillation does to a model. On the governance side it covers guardrail policy types, model cards, identity-based against resource-based policies, key policies, and what CloudTrail records by default. You get 95 minutes for 65 questions and need 80 percent to pass. No question depends on memorising a service quota, a price or a context window size, because those change; the questions are about mechanism and order, which do not. Every option carries its own explanation and every question links to the AWS documentation page that supports the answer.

Exam Day: Full-Length Timed Simulation

65 questions · 90 minutes · pass mark 85%

Sixty-five questions in 90 minutes, the exact length and timing of the real AIF-C01 exam. Every question here is a full scenario with several constraints stated at once, and the domains are mixed the way they are on exam day: a security question may turn on a retrieval design detail, a generative AI question may turn on a responsible AI constraint. The stems are the longest in this practice set by design, because reading speed under time pressure is part of what the real exam measures. Two questions simulate the ordering and matching formats AWS uses. The pass mark is 85 percent rather than the 80 percent of the earlier tests, and deliberately well above what the real exam requires: passing this test in one uninterrupted sitting is the signal that you are ready to book. Take it that way - no pausing, no looking anything up, no second attempt on the same day. Afterwards, read every explanation, including those for the questions you answered correctly, and look at your score by domain before deciding what to revise. Every question links to the AWS documentation behind its answer.

What you will practice

  • Sit five full-length AIF-C01 practice exams of 65 questions in 90 minutes, matching the real exam exactly.
  • Tell apart the AWS AI services that are easiest to confuse, and know which one each kind of task calls for.
  • Decide between RAG, fine-tuning, continued pre-training and distillation from cost, latency and data constraints.
  • Read your score by exam domain after every attempt and know which of the five domains to revise next.
  • Understand why every answer is right or wrong, with a per-option explanation and a link to the AWS documentation.
  • Practise the ordering and matching question formats AWS uses and that most practice test courses leave out.

Exam domains covered

Applications of Foundation Models28%
Fundamentals of GenAI25%
Fundamentals of AI and ML20%
Guidelines for Responsible AI14%
Security, Compliance, and Governance for AI Solutions14%

Who is this for

Anyone preparing for the AWS Certified AI Practitioner (AIF-C01) exam. That includes developers and IT professionals moving toward AI and machine learning work, cloud engineers and administrators adding AI to what they already know, and students and career changers taking their first AWS certification. It equally includes the non-technical roles the exam was written for: product managers, business analysts, sales and marketing staff, project managers, and legal, risk and compliance professionals who have to reason about AI systems without building them.

AIF-C01 has no retired predecessor - it is the first version of this certification, launched in 2024 - but its content has changed substantially. The exam guide was reissued on 30 April 2026 and now covers agentic AI, the Model Context Protocol, context engineering, token-based pricing, prompt management and hallucination grounding, while dropping several services that used to be in scope. Practice material written before that date tests a syllabus that no longer exists. This practice set is built on the 30 April 2026 guide.

What this is: 325 exam-style questions across five timed tests, each answer option explained, each question sourced to the AWS documentation. What it is not: a video course, a substitute for the AWS learning material, a guarantee of a pass, and it contains no real exam questions.

Requirements

  • No AWS experience is required. AIF-C01 is a foundational certification and the questions assume no coding.
  • Some prior study of the AIF-C01 material. These are practice exams, not a course: they test and explain, they do not teach the syllabus from scratch. Use them alongside the free AWS Skill Builder exam prep course and the AWS documentation, not instead of them.
  • A willingness to read the explanations. Most of the value here is in the reasoning after the answer, not in the score.

Free exam guide

Not sure what to expect? Read how the exam is built, what it covers and how to prepare.