Certification guide
Microsoft AI-900 is the fundamentals-level certification that validates a working, conceptual understanding of artificial intelligence and machine learning on Microsoft Azure. Passing it earns the Microsoft Certified: Azure AI Fundamentals credential. It is intended for anyone – technical or non-technical – who needs to demonstrate AI literacy in the Microsoft ecosystem, from developers stepping into AI work, to product managers, analysts, sales engineers, and students planning to move on toward the role-based Azure AI developer or Azure Data Scientist tracks.
The exam does not require Python coding, deep learning experience, or statistics. It expects you to recognize AI workload categories, match Azure AI services to real-world scenarios, apply Microsoft’s Responsible AI principles, and – since the January 2026 refresh – handle a substantial block of questions on generative AI, Azure OpenAI Service, and Microsoft Copilot products. Materials written before that refresh miss a large portion of what is now on the exam, so up-to-date preparation matters more here than on most fundamentals exams.
Artificial intelligence workloads and considerations
Domain 1 · 15-20%Recognize the common AI workload types – prediction and forecasting, anomaly detection, computer vision, natural language processing, document intelligence, knowledge mining, and generative AI – and match each to real business scenarios. This domain also covers Microsoft’s six Responsible AI principles (fairness, reliability and safety, privacy and security, inclusiveness, transparency, and accountability) and how each shows up in the design and operation of a system. Expect scenario stems where the right answer hinges on which principle is most directly at risk.
Fundamental principles of machine learning on Azure
Domain 2 · 15-20%Distinguish regression, classification, and clustering; understand features, labels, training and validation datasets, and the difference between supervised and unsupervised learning. Interpret common evaluation metrics – accuracy, precision, recall, F1, ROC/AUC for classification; MAE, RMSE, and R-squared for regression – and read a confusion matrix correctly. On the Azure side, know when to reach for Azure Machine Learning studio, Designer’s drag-and-drop pipelines, Automated ML, and notebooks. This domain rewards clean vocabulary more than math.
Computer vision workloads on Azure
Domain 3 · 15-20%Separate the core vision tasks – image classification, object detection, semantic segmentation, OCR, face detection versus face verification and recognition – and know the output each produces. Match the Azure services: Azure AI Vision for prebuilt image analysis and OCR, Azure AI Custom Vision for training on your own labeled images, Azure AI Face for face-specific workloads, and Azure AI Document Intelligence for structured extraction from invoices, receipts, and forms. Many stems test which service to pick when several sound close.
Natural language processing workloads on Azure
Domain 4 · 15-20%Cover the Azure AI Language features – sentiment analysis, key phrase extraction, named entity recognition, PII detection, language detection, custom question answering, and conversational language understanding for intent and entity extraction – plus Azure AI Translator for text and speech translation and Azure AI Speech for text-to-speech, speech-to-text, and speaker recognition. The exam frequently pairs a business scenario (categorize reviews, redact identifiers, route by language) with the smallest service that solves it.
Generative AI workloads on Azure
Domain 5 · 20-25%The heaviest domain on the current exam. Understand what generative AI is and how large language models work at a conceptual level – tokens, prompts, embeddings, temperature, and the transformer idea. Know Azure OpenAI Service and its GPT and DALL-E model families; know grounding and retrieval-augmented generation as the standard defense against hallucination; and know the Microsoft Copilot lineup – Microsoft 365 Copilot, Copilot Studio, GitHub Copilot, and Copilot in Windows – plus Azure AI Foundry as the platform for building and evaluating generative applications. Content filters and responsible generative AI practices are consistently tested.
The domain weights matter because the exam is short. On a 40-question form, a shaky Domain 5 costs you eight or nine questions all by itself, so generative AI is the study block to invest in first. The other four domains sit within a few percentage points of each other, and questions frequently cross domain boundaries in scenario stems – a policy chatbot touches both NLP and generative AI, a defect-detection kiosk touches vision and responsible AI. Practice questions written to the current outline are the fastest way to close the gap between “I read the module” and “I recognize the stem pattern.”
Your Certifym.net practice exam below mirrors the current domain weights and question types. The 70% pass mark is an honest raw-score equivalent to the Microsoft 700-of-1000 passing threshold – it holds you to the same bar you will face on exam day, not luck in the heavy ones. Treat a first-attempt score at or above 80% on this practice set as a strong signal that you are ready to book.
Microsoft Azure AI Fundamentals - Practice Exam
Comprehensive practice bank covering all five AI-900 domains: AI workloads and considerations, machine learning principles, computer vision, natural language processing, and generative AI on Microsoft Azure.…
Subscribe to startFrequently asked questions about AI-900
What is the Microsoft AI-900 exam?
AI-900 is Microsoft’s fundamentals-level certification exam covering artificial intelligence and machine learning on Microsoft Azure. Passing it earns the Microsoft Certified: Azure AI Fundamentals credential, which validates a working, conceptual understanding of AI workloads, Azure AI services, Responsible AI, and generative AI.
How many questions are on the AI-900 exam and how long is it?
AI-900 forms typically run about 40 to 60 items in roughly 45 to 60 minutes. Because the exam is short, each domain carries real weight – on a 40-question form, a shaky Domain 5 costs you eight or nine questions all by itself.
What is the passing score for AI-900?
Microsoft’s passing threshold is 700 out of 1000 on its scaled scoring system. The Certifym practice exam on this page uses a 70% pass mark as an honest raw-score equivalent of that 700-of-1000 threshold.
Are there prerequisites for AI-900?
None. There is no prerequisite certification, and the exam does not require Python coding, deep learning experience, or statistics. It is designed for technical and non-technical candidates alike – developers stepping into AI work, product managers, analysts, sales engineers, and students.
How hard is the AI-900 exam?
It is a fundamentals exam, so it tests recognition and vocabulary rather than implementation – you are asked to recognize AI workload categories, match Azure AI services to real-world scenarios, and apply Microsoft’s Responsible AI principles. The difficulty sits in service selection: many stems present several Azure services that sound close, and only one fits the scenario.
What domains does AI-900 cover and how are they weighted?
Five domains: artificial intelligence workloads and considerations (15-20%), fundamental principles of machine learning on Azure (15-20%), computer vision workloads on Azure (15-20%), natural language processing workloads on Azure (15-20%), and generative AI workloads on Azure (20-25%). Generative AI is the heaviest domain on the current exam and the study block to invest in first.
Does the Azure AI Fundamentals certification expire?
No renewal is required. Unlike Microsoft’s role-based associate and expert credentials, the Azure AI Fundamentals certification does not carry an annual renewal requirement.
What changed in the January 2026 AI-900 refresh?
The refresh added a substantial block of questions on generative AI, Azure OpenAI Service, and Microsoft Copilot products, and generative AI now stands as its own 20-25% domain. Materials written before that refresh miss a large portion of what is now on the exam, so up-to-date preparation matters more here than on most fundamentals exams.
How do I prepare for AI-900?
Study to the current exam outline, with generative AI first, and then drill scenario stems. Practice questions written to the current outline are the fastest way to close the gap between “I read the module” and “I recognize the stem pattern.” Treat a first-attempt score at or above 80% on the Certifym practice set as a strong signal that you are ready to book.
How does AI-900 compare with the role-based Azure AI certifications?
AI-900 is conceptual and requires no coding; it is the literacy credential candidates earn before moving on toward the Azure AI Engineer or Azure Data Scientist role-based tracks. Those role-based exams assume hands-on implementation work, while AI-900 asks you to recognize workload types and pick the right Azure AI service for a scenario.
Trademark notice & independence. Certifym.net is operated by Certifym Exam Services, LLC and is not affiliated with, endorsed by, or sponsored by Microsoft Corporation. Microsoft, Azure, Microsoft Certified, and Microsoft 365 Copilot are trademarks or registered trademarks of Microsoft Corporation. AI-900 and the Microsoft Certified: Azure AI Fundamentals credential are administered solely by Microsoft.
All practice questions, explanations, and study materials on Certifym.net are original content authored to align with the publicly documented AI-900 exam skills outline. Certifym.net does not reproduce, mirror, or paraphrase actual live exam questions, and no material on this site is drawn from Microsoft’s proprietary item bank.