AI-901 in 2026: Microsoft's Rebuilt Azure AI Fundamentals Exam
Last updated: July 3, 2026. Verified against the official Microsoft Learn AI-901 exam page, the AI-901 study guide, the Azure AI Fundamentals certification page, and Liberty Munson's April 15, 2026 announcement on the Microsoft Tech Community.
AI-901 is not a brand-new certification. It is the rebuilt exam behind the same credential you may already know: Microsoft Certified: Azure AI Fundamentals. On April 15, 2026, Microsoft announced that Exam AI-901: Microsoft Azure AI Fundamentals would replace Exam AI-900, which was retired on June 30, 2026. The certification itself was not retired — only the exam required to earn it changed. If you hold the Azure AI Fundamentals certification from AI-900, it stays valid.
The rebuild is more than a renumbering. AI-900 asked "What is AI?" and tested whether you could describe AI/ML concepts and identify when to use Azure AI services. AI-901 asks "How do I build an AI app using Microsoft Foundry?" and tests whether you can implement AI solutions with the Foundry portal, SDK, and Tools. The new exam also requires basic Python — a first for an Azure fundamentals AI exam.
AI-901 at a Glance
The official Exam AI-901 page and the AI-901 study guide publish these facts directly.
| Item | Verified detail |
|---|---|
| Exam code | AI-901 |
| Exam name | Microsoft Azure AI Fundamentals |
| Certification earned | Microsoft Certified: Azure AI Fundamentals |
| Job role (per certification page) | AI Engineer |
| Passing score | 700 (on a 1,000-point scale) |
| Language | English (updated April 15, 2026) |
| Retirement date | None |
| Skills measured domains | 2 |
| Domain 1 | Identify AI concepts and capabilities (40–45%) |
| Domain 2 | Implement AI solutions by using Microsoft Foundry (55–60%) |
| Required coding | Basic Python syntax and programming concepts |
| Instructor-led course | Course AI-901T00-A: Introduction to AI in Azure (1 day, beginner) |
Two items competitors often state as fact that the official Microsoft Learn page does not publish: a fixed question count (often cited as 40–60) and a fixed time limit (often cited as 45 minutes). The AI-901 exam page does not list either number. Those figures are inherited from AI-900 and may or may not match the generally available form. Treat them as estimates, not verified rules, and confirm the format when you schedule.
The Verified Skills Measured on AI-901
The AI-901 study guide lists the skills measured as of April 15, 2026. Microsoft notes that the bullets under each skill illustrate how the skill is assessed and that related topics may also appear, and that most questions cover generally available features but commonly used preview features may also appear. You should be familiar with REST APIs, SDKs, and CLIs.
Domain 1: Identify AI Concepts and Capabilities (40–45%)
This domain tests whether you can recognize AI workloads, model choices, and responsible AI principles. It is conceptual, but the topics are narrower and more modern than AI-900's five-domain spread.
Describe principles of responsible AI
- Fairness, reliability and safety, privacy and security, inclusiveness, transparency, and accountability — the six responsible AI principles Microsoft publishes for Azure AI.
Identify AI model components and configurations
- Describe how generative AI models work.
- Identify an appropriate AI model based on capabilities.
- Identify appropriate model deployment options and configuration parameters.
Identify AI workloads
- Identify scenarios for common AI workloads: generative and agentic AI, text analysis, speech, computer vision, and information extraction.
- Describe common text analysis techniques: keyword extraction, entity detection, sentiment analysis, and summarization.
- Identify features and capabilities of speech recognition and speech synthesis.
- Identify features and capabilities of computer vision and image-generation models.
- Identify techniques to extract information from text, images, audio, and videos.
Domain 2: Implement AI Solutions by Using Microsoft Foundry (55–60%)
This is the domain that makes AI-901 a different exam from AI-900. It tests hands-on implementation inside Microsoft Foundry — the unified Azure platform for building, deploying, and managing AI apps and agents. The study guide breaks it into four sub-domains.
Implement generative AI apps and agents by using Foundry
- Create effective system and user prompts for generative AI models.
- Deploy a model and interact with it in the Foundry portal.
- Create a lightweight chat client application by using the Foundry SDK.
- Create and test a single-agent solution in the Foundry portal.
- Create a lightweight client application for an agent.
Implement AI solutions for text and speech by using Foundry
- Build a lightweight application that includes text analysis.
- Respond to spoken prompts by using a deployed multimodal model.
- Build a lightweight application by using Azure Speech in Foundry Tools.
Implement AI solutions with computer vision and image-generation capabilities by using Foundry
- Interpret visual input in prompts by using a deployed multimodal model.
- Create new visual outputs by using generative models.
- Build a lightweight application that includes vision capabilities.
Implement AI solutions for information extraction by using Foundry
- Extract information from documents and forms by using Azure Content Understanding in Foundry Tools.
- Extract information from images by using Content Understanding.
- Extract information from audio and video by using Content Understanding.
- Build a lightweight application with information extraction capabilities by using Content Understanding.
The weighting tells you where to spend your prep time. Because Domain 2 is 55–60% and Domain 1 is 40–45%, roughly six of every ten questions touch Foundry implementation. If you only study concepts and skip hands-on Foundry work, you are leaving more than half the exam score on the table.
How AI-901 Differs from AI-900
Liberty Munson's announcement on the Microsoft Tech Community lays out the shift directly. The table below mirrors Microsoft's published comparison.
| Area | AI-900 (retired June 30, 2026) | AI-901 (current) |
|---|---|---|
| Audience | Nontechnical or technical beginners | Technical beginners planning to build AI solutions |
| Skills tested | Understanding AI/ML concepts; identifying when to use Azure AI services | Understanding AI concepts; implementing AI solutions with Microsoft Foundry |
| Required coding | None | Basic Python syntax and programming concepts |
| Knowledge level | Introductory, conceptual understanding of Azure AI services | Foundational Azure resources plus Foundry implementation, provisioning, and modern AI tooling |
| Focus | "What is AI?" | "How do I build an AI app by using Foundry?" |
| Domains | 5 (workloads, ML, vision, NLP, generative AI) | 2 (AI concepts, Foundry implementation) |
| Largest domain | Generative AI workloads (20–25%) | Foundry implementation (55–60%) |
What is de-emphasized in AI-901 compared with AI-900: traditional ML subtopics as named exam areas (regression, classification, clustering), Azure Machine Learning Studio pipelines and automated ML, the Azure Bot Service and conversational AI framing (replaced by agentic AI), and individual Azure Cognitive Services naming (collapsed into Microsoft Foundry services). What is new: AI agents and multi-step agentic workflows, Foundry as the unified hub, multimodal models including image generation, Azure Content Understanding for information extraction, and prompt engineering as a tested skill.
If you already passed AI-900, you do not need to take AI-901. The certification page and the announcement both state existing holders need take no action. If you want to showcase the broader Foundry and agentic AI skills, you may optionally take AI-901.
Exam Format, Pricing, and Scheduling
A few format facts are verified directly from the official AI-901 exam page:
- Passing score: 700. Microsoft's exam scoring documentation explains that a score of 700 or greater is required to pass.
- Language: English. The English version was updated April 15, 2026. Microsoft notes that localized versions, when available, are updated approximately eight weeks after the English version.
- Retirement date: None.
- Pricing: The exam page states "Price based on the country or region in which the exam is proctored" and directs you to confirm exact pricing with the exam provider before registering. Microsoft does not publish a single global fee on the AI-901 page.
- Registration: Microsoft recommends registering with a personal Microsoft account (not a work or school AAD account) so your exam records stay with you.
A few format details competitors often cite that Microsoft does not publish on the AI-901 page: a fixed question count, a fixed time limit, and a 99 USD price. Those numbers are inherited from AI-900 and third-party reporting. The AI-901 study guide does note that the practice assessment is not yet available and is usually released within eight weeks of the exam becoming generally available.
Beta exam context (for reference)
AI-901 launched as a beta. Per Liberty Munson's April 15, 2026 announcement, the first 300 people who took the beta on or before May 6, 2026 could get 80% off using code AI901Medford (not available in Turkey, Pakistan, India, or China). Beta exam rescoring began when the exam went live, with final results released approximately 10 days later. The credential earned from the beta is identical to the generally available version. The beta period has now closed; if you are scheduling today, you register for the generally available exam at standard pricing.
Who Should Take AI-901
The official AI-901 exam page describes the candidate as someone "at the beginning of your career in AI solution development." Prerequisites it lists:
- Conceptual knowledge of AI solutions in Azure.
- Foundational technical skills to work with Azure AI solutions.
- Knowledge of Python coding syntax and programming techniques.
- Familiarity with Azure resources.
The Python requirement is the biggest change from AI-900. You do not need to be a software engineer, but you should be comfortable reading short Python snippets and understanding syntax, variables, function calls, and basic control flow. If you have never written Python, plan a few days of Python basics before starting the Foundry implementation modules.
AI-901 is a fundamentals exam. It is not a mandatory prerequisite for the associate-level Azure AI certifications, but it is a useful on-ramp if you plan to pursue role-based Azure AI Engineer credentials later. For a broader view of where AI-901 sits among Microsoft fundamentals, read our AZ-900 vs AI-900 vs SC-900 comparison and our Azure AI Fundamentals AI-900 exam guide (the legacy exam content is still useful background for Domain 1).
How to Prepare: Microsoft's Recommended Resources
The AI-901 exam page and the AI-901 study guide point to two official preparation paths.
Self-paced learning paths. Microsoft publishes two learning paths for AI-901:
- AI concepts for developers and technology professionals (6 modules).
- Get started with AI applications and agents on Azure (6 modules).
Together the two paths cover 12 modules spanning both measured domains. Start with the AI concepts path to build Domain 1 vocabulary, then move into the apps and agents path for Domain 2.
Instructor-led course. Course AI-901T00-A: Introduction to AI in Azure is a one-day beginner-level course that introduces AI concepts and the Azure services used to build AI solutions. Microsoft describes it as teaching "a mix of AI concepts and technology skills that are considered foundational to a successful career implementing AI solutions on Microsoft Azure."
Hands-on Foundry time. Because 55–60% of the exam tests Foundry implementation, reading alone is not enough. Open a Foundry portal, deploy a model, write system and user prompts, build a lightweight chat client with the Foundry SDK, create and test a single agent, and use Azure Content Understanding on a document and an image. The exam expects you to be familiar with REST APIs, SDKs, and CLIs, so do not skip the SDK exercises.
A Four-Week AI-901 Study Plan
Adjust the length if your test is closer or further out, but keep the order: concepts first, Foundry implementation second, full practice last.
| Week | Main job | What to do |
|---|---|---|
| 1 | Build concept vocabulary | Complete the AI concepts learning path (6 modules). Learn the six responsible AI principles, generative AI model basics, model selection, and the five workload families (generative/agentic, text, speech, vision, information extraction). Start a Python basics refresher if needed. |
| 2 | Start Foundry hands-on | Open the Foundry portal. Deploy a model, write system and user prompts, and interact with the model. Build a lightweight chat client with the Foundry SDK. Begin the apps and agents learning path. |
| 3 | Cover the rest of Domain 2 | Build text analysis, speech, vision, and image-generation lightweight apps. Use Azure Content Understanding on documents, images, and audio/video. Create and test a single agent in the Foundry portal and a client app for it. |
| 4 | Practice and sharpen | Take full mixed practice sets. Track misses by domain and sub-skill. Revisit study-guide sections only where the miss log points. Run the exam sandbox so the UI is familiar on test day. Confirm scheduling, ID, and language. |
For most learners, three short concept sessions, three Foundry hands-on sessions, and one mixed timed practice set per week is a sustainable rhythm. The score gain comes from understanding why a Foundry implementation choice is right or wrong, not from memorizing service names.
Test-Day Checklist
- Confirm your exam is AI-901 (not AI-900, which is retired) when you schedule.
- Register with a personal Microsoft account so your records stay portable.
- Confirm the exam language and pricing shown at checkout for your country/region.
- Run the system check and exam sandbox before test day so the environment is familiar.
- Bring the ID you used at registration, with the exact name match.
- Answer every question — a blank cannot help you, and fundamentals exams do not publish a separate wrong-answer penalty.
- For Foundry implementation questions, mentally visualize the portal, SDK, and Tools flow rather than guessing from service names.
- For responsible AI questions, match the scenario to the specific principle (fairness, reliability and safety, privacy and security, inclusiveness, transparency, accountability) rather than choosing a generic "ethical" option.
Best Next Step
AI-901 rewards a different kind of preparation than AI-900 did. Conceptual reading earns you Domain 1 (40–45%). Hands-on Foundry work earns you Domain 2 (55–60%) and the Python literacy to follow the SDK examples. Treat the exam as a build-and-implement test, not a describe-and-recognize test, and your prep becomes focused instead of scattered.
Buy on Amazon