Skip to content
50% offFounding price until 1 DecemberSee prices
Square1 Ai
Get started
All certifications

Microsoft · AI-901 · Fundamentals

Microsoft Azure AI Fundamentals

Microsoft's AI fundamentals for people starting to build: responsible AI, how generative models work, and building small apps and agents in Microsoft Foundry.

Free practice: 195 questions written to the official exam guide, with the diagnostic, practice papers, timed mocks, a study plan and an explanation on every answer.

Official exam page

Find out where you stand on AI-901

20 questions drawn in the exam’s own domain weights. You get a report straight away: your score against the pass mark, where the points went, and what to study first.

Questions
20
Time
about 20 minutes
Cost
free

Your email

Your answers save as you go, and the report stays in your record. A new email needs no code.

· your answers stay on this device and you can save the report at the end.

The course

The diagnostic shows where you stand. Then you learn the syllabus chapter by chapter, answer that chapter’s questions, and move on. When every chapter is mastered, you sit full timed mocks in the real exam’s shape.

12 chapters, about 9 hours in all. Each is a 45-minute study session with quick checks, then a 10-question quiz on that chapter.

Identify AI concepts and capabilities42% of the exam

  1. Responsible AI: the six principles45 min
  2. How generative AI models work45 min
  3. Choosing, deploying and configuring a model45 min
  4. Recognising AI workloads45 min
  5. Text analysis and speech: the concepts45 min
  6. Vision, image generation and information extraction: the concepts45 min

Implement AI solutions by using Microsoft Foundry58% of the exam

  1. Prompts and the Foundry portal45 min
  2. A chat client with the Foundry SDK45 min
  3. Building a single agent in Foundry45 min
  4. Text and speech solutions in Foundry45 min
  5. Vision and image generation in Foundry45 min
  6. Information extraction with Content Understanding45 min

About the Microsoft Azure AI Fundamentals (AI-901) exam

AI-901 is the exam for Microsoft Certified: Azure AI Fundamentals. It replaced AI-900, which Microsoft retired on 30 June 2026. The certification kept its name; the exam changed character. AI-900 asked whether you could name the AI workloads and the Azure service behind each one. AI-901 asks whether you can start building with them.

It is aimed at people at the beginning of a career in AI solution development. Microsoft expects Python syntax, familiarity with Azure resources, and comfort with REST APIs, SDKs and command-line tools. The exam has two parts: the concepts (responsible AI, how generative models work and how to configure them, and the common AI workloads) and implementation in Microsoft Foundry, which is the larger part: prompting and deploying models, a small chat client and a single agent, text and speech, vision and image generation, and information extraction with Content Understanding.

At the time of writing Microsoft lists no learning path or instructor-led course for AI-901, only the study guide and a practice assessment on AI Skills Navigator. That makes a question bank written against the study guide, with an explanation on every answer, more useful here than for most fundamentals exams.

The exam at a glance

Questions
Not published on the exam page at the time of writing; Microsoft exams mix multiple choice with other item types
Time
Not published on the exam page at the time of writing; check it when you book
Pass mark
700 out of 1,000
Exam fee
Priced by the country you sit it in; check the exam page when you book
Valid for
Does not expire (Microsoft Fundamentals certifications have no renewal)
Where you sit it
Pearson VUE test centre or online proctored; practice assessment on AI Skills Navigator
Prerequisites
None required. Microsoft expects Python syntax and familiarity with Azure resources

As published in the vendor’s exam guide at the time of writing. Fees vary by country and change; the vendor’s page is the source of truth.

The fee and how to sit it in your country: Sri Lanka · India · Pakistan · Bangladesh · Nepal · Australia · New Zealand · Philippines · Indonesia · Vietnam

What is on the exam, domain by domain

  1. Identify AI concepts and capabilities

    About 40 to 45 percent. Microsoft's six responsible AI principles; how generative models work, choosing a model for a capability, deployment options and parameters such as temperature and max tokens; and recognising the workload in a scenario: generative and agentic AI, text analysis, speech, computer vision and information extraction.

    42% of the paper

  2. Implement AI solutions by using Microsoft Foundry

    About 55 to 60 percent, the larger part. System and user prompts; deploying a model and using it in the Foundry portal; a small chat client with the Foundry SDK; a single agent built and tested in the portal and called from code; text analysis and spoken prompts; Azure Speech; images in prompts, image generation and vision; and extracting information from documents, images, audio and video with Content Understanding.

    58% of the paper

How to prepare

  1. Read the AI-901 study guide first and ignore older AI-900 material. Half of what AI-900 tested is gone, and the Foundry half of AI-901 did not exist in it.
  2. Get an Azure account and do the Foundry quickstarts: deploy a chat model, call it from a short Python script, then build one agent with a tool and call it from code. The implementation questions are much easier once you have written those few lines yourself.
  3. Learn the six responsible AI principles by name and practise telling them apart in a scenario. They are asked directly, and fairness, inclusiveness and transparency are easy to confuse.
  4. For each workload (text, speech, vision, image generation, information extraction), know when a multimodal model is enough and when a dedicated Foundry tool such as Azure Speech or Content Understanding is the better fit.
  5. Sit a diagnostic before you study, then practise in the exam's weights until you clear your target twice running. Book the exam after that, not before.

Sample questions

Three questions in the exam’s style, written by Square 1 against the published domains. They are ours, not the vendor's. The answer and the reasoning are under each one; the practice papers on this page draw from a bank written the same way, with an explanation on every question.

  1. Question 1 · Identify AI concepts and capabilities

    A bank's loan assistant gives applicants a decision but never says which factors drove it, and applicants cannot find out. Which responsible AI principle does this most directly fall short of?

    1. Transparency
    2. Inclusiveness
    3. Reliability and safety
    4. Privacy and security

    Answer: A. Transparency means people can understand how an AI system reaches its outputs and what it is for. Inclusiveness is about designing for the full range of people who use a system; it is a different gap from an unexplained decision.

  2. Question 2 · Identify AI concepts and capabilities

    A developer's deployed chat model gives noticeably different wording each time it answers the same support question, and the team wants answers that are as consistent as possible. Which parameter should the developer lower?

    1. Temperature
    2. Max tokens
    3. Number of past messages sent
    4. Frequency penalty

    Answer: A. Temperature controls how much randomness the model uses when choosing each token; lowering it makes outputs more deterministic. Max tokens only caps the length of a response, so lowering it shortens answers without making them consistent.

  3. Question 3 · Implement AI solutions by using Microsoft Foundry

    A developer wants a deployed multimodal model to describe what is in a photo a user uploads to a chat app. How should the photo reach the model?

    1. As image content in the user message, by URL or as encoded data
    2. As a separate fine-tuning job run before the conversation starts
    3. As a file placed in the project's storage for the model to scan
    4. As text in the system message, converted from the image first

    Answer: A. A multimodal chat model accepts images as part of a message's content, either as a URL or as base64-encoded data, alongside any text. Fine-tuning changes a model's behaviour over many examples; it is not how a model sees one image in one conversation.

Questions people ask before they book

Is AI-901 the same as AI-900?
No. AI-900 was retired on 30 June 2026 and AI-901 replaced it for the same certification, Azure AI Fundamentals. AI-901 is more technical: over half of it is about building small solutions in Microsoft Foundry, and Microsoft expects basic Python.
I already passed AI-900. Do I need AI-901?
No. The Azure AI Fundamentals certification is not retiring, and a certification earned through AI-900 stays valid. AI-901 is for people earning it now.
Do I need to be able to code?
You need to read and write basic Python and follow SDK examples. The exam does not ask you to write long programs, but the implementation questions assume you know what a short chat client or agent client does line by line.
How long should I study?
Plan for 25 to 40 hours if you are new to Azure, less if you already build with Python. Put most of it into the Foundry half, which is the larger share of the exam, and into hands-on quickstarts.
AI-901 or the AWS AI Practitioner first?
Pick the cloud your employer or target employers use. The AWS AI Practitioner is concept-first and needs no code; AI-901 expects you to build a little. If you want the hands-on start, AI-901 is the better fit.