Is Azure AI-900 a Proctored Exam?
Yes. Azure AI-900 is a fully proctored exam, available at Pearson VUE testing centers worldwide or via online proctoring through Pearson VUE's OnVUE system. Online proctored exams require a webcam, microphone, and clean desk in a private room. You must present a valid government-issued ID before the exam begins.
This is fundamentally different from Google AI Essentials, which is a self-paced course with unproctored quizzes. Azure AI-900 is a traditional certification exam: you book a time slot in advance, the clock runs continuously from the moment you begin, and your score is reported immediately after finishing.
Full exam details are at learn.microsoft.com/en-us/credentials/certifications/azure-ai-fundamentals.
What Are the Azure AI-900 Exam Details?
- Exam fee: $99 USD (pricing varies by country)
- Number of questions: 40 to 60 questions
- Testing time: 45 minutes (65 minutes total seat time, which includes check-in, tutorial, and survey)
- Passing score: 700 on a scale of 100–1,000
- Question types: Multiple choice, multiple response, drag-and-drop, and ordering questions
- Retake policy: If you fail, you must wait at least 24 hours before retaking. After a second failure, the wait period extends to 14 days. The full $99 fee applies to each retake attempt.
- Certification validity: Microsoft certifications at the Fundamentals level do not expire.
The 45-minute testing time is the shortest of the three certifications in this guide—45 minutes for up to 60 questions means as little as 45 seconds per question on average. Timed practice is essential.
What Are the Five Azure AI-900 Skill Areas?
The exam is organized into five skill areas with percentage weights that reflect how many scored questions cover each area:
- Describe AI workloads and considerations — 15–20% — AI use case types, responsible AI principles, AI considerations (fairness, reliability, privacy, transparency, inclusiveness, accountability)
- Describe fundamental principles of machine learning on Azure — 20–25% — ML types, pipeline stages, evaluation metrics, Azure Machine Learning service, AutoML
- Describe features of computer vision workloads on Azure — 15–20% — Azure AI Vision, Custom Vision, Face, Video Indexer, Document Intelligence
- Describe features of natural language processing (NLP) workloads on Azure — 15–20% — Azure AI Language, Translator, Speech, Bot Service, conversational language understanding
- Describe features of generative AI workloads on Azure — 20–25% — Foundation models, Azure OpenAI Service, prompt engineering, responsible generative AI, Microsoft Copilot
What Question Types Appear on Azure AI-900?
Azure exams use a wider variety of question formats than AWS exams at the foundational level:
- Multiple choice: Select one correct answer from four options. The most common question type.
- Multiple response: Select all correct answers (typically two or three) from a set of five or more options. Partial credit is not given—you must select all correct answers and no incorrect ones.
- Drag-and-drop: Match Azure AI services to use cases, or arrange steps of a process in the correct order. These require knowing service names precisely—typos do not exist in drag-and-drop, but confusing similar service names (Azure AI Language vs. Azure AI Translator vs. Azure AI Speech) costs points.
- Ordering questions: Arrange steps in a process (like the ML pipeline) in the correct sequence.
Drag-and-drop and ordering questions are unique to Microsoft exams and do not appear in AWS AI Practitioner or Google AI Essentials. Practicing with materials that include these question types is important—seeing them for the first time during the exam costs precious seconds from the 45-minute limit.