Azure AI Fundamentals (AI-901)

Microsoft

Complete guide to passing the Azure AI Fundamentals (AI-901) exam on your first attempt.

MediumHigh Search Volume
Key Information at a Glance
Cost

$99

Pass Rate

Not published

Validity

Does not expire

Region

Global

Provider

Microsoft

Salary Impact

$61k-$131k (BLS 2024 medians)

Are you ready for Azure AI Fundamentals (AI-901)?

Loading quiz...

Complete Overview

Exam AI-901 is the single exam required for the Microsoft Certified: Azure AI Fundamentals certification, and it costs $99 USD in the United States, with the price set by the country or region where the exam is proctored. It is aimed at people at the beginning of a career in AI solution development, and Microsoft states that candidates should have conceptual knowledge of AI solutions in Azure, knowledge of Python coding syntax and programming techniques, and familiarity with Azure resources. AI-901 replaced exam AI-900, which Microsoft retired on 30 June 2026. The English language version of AI-901 was updated on 15 April 2026, and the exam page lists no retirement date. Anyone who passed AI-900 before 30 June 2026 already holds the same Azure AI Fundamentals certification and does not need to sit AI-901. Microsoft publishes two skill groups for the exam. Identify AI concepts and capabilities carries 40 to 45 percent of the scored content. Implement AI solutions by using Microsoft Foundry carries 55 to 60 percent. That second group is the largest change from AI-900: the exam now asks candidates to deploy a model in the Foundry portal, write prompts, build a lightweight chat client with the Foundry SDK, create and test a single-agent solution, and extract information from documents, images, audio, and video using Azure AI Content Understanding. The scoring scale runs from 1 to 1,000 and the passing score is 700. Microsoft describes this as a scaled score that may not equal 70 percent of the available points. Microsoft publishes 45 minutes of exam duration and 65 minutes of seat time for all Fundamentals exams, and Certiport's published exam length list gives AI-901 a range of 42 to 46 questions. Five minutes of break time is built into the exam clock, and the clock keeps running during a break. The exam is delivered by Pearson VUE at test centres and through OnVUE online proctoring, and by Certiport for students and educators. Thirteen languages are offered: English, Arabic (Saudi Arabia), Chinese (Simplified), Chinese (Traditional), French, German, Indonesian, Italian, Japanese, Korean, Portuguese (Brazil), Russian, and Spanish. Candidates taking the exam in English who are not native English speakers can request 30 extra minutes. Unlike associate and expert exams, Fundamentals exams do not give candidates access to Microsoft Learn during the test. Microsoft Fundamentals certifications do not expire, so passing AI-901 once earns a credential that never needs renewal, in contrast to associate, expert, and specialty certifications, which expire annually and are renewed through a free online assessment. Microsoft does not publish a pass rate for AI-901 or any other certification exam. If you fail, you wait 24 hours before the second attempt and 14 days before each attempt after that, up to five attempts in a 12-month period measured from the first attempt.

Why Get Azure AI Fundamentals (AI-901) Certified?

One exam of 45 minutes earns a full Microsoft certification, compared with 100 to 120 minutes for associate and expert role-based exams.

The $99 USD fee is the lowest tier Microsoft charges; associate and expert role-based exams are priced higher.

Microsoft Fundamentals certifications do not expire, so the $99 is a one-time cost with no annual renewal assessment.

55 to 60 percent of the scored content is hands-on Microsoft Foundry work, so the certificate signals practical build experience rather than terminology recall alone.

The exam is available in 13 languages, so candidates outside English-speaking markets can test in their own language.

AI-901 is the current code with no retirement date listed, while AI-900 stopped being available on 30 June 2026.

Passing score is a fixed 700 on a 1 to 1,000 scale, published openly by Microsoft rather than hidden behind a variable cut.

Exam Format & Structure

Duration

45 minutes of exam time, 65 minutes of seat time

Questions

42 to 46 questions, per Certiport's published exam length list; Microsoft states most of its certification exams contain 40 to 60 questions and does not publish a fixed count for AI-901

Passing Score

700 on a scale of 1 to 1,000

Question Types

  • Multiple choice
  • Multiple response
  • Drag and drop
  • Build list
  • Hot area
  • Active screen

Delivery Method

Pearson VUE test centre or OnVUE online proctored exam; Certiport delivers the exam for students and educators

  • Microsoft builds five minutes of break time into the exam clock and removes questions to make room for it. The clock keeps running while you are on break.
  • Once you start a break you cannot return to any question you have already seen, including questions left unanswered or marked for review.
  • Access to Microsoft Learn during the exam is available on associate and expert role-based exams only. Fundamentals exams do not have it.
  • Microsoft does not identify which question types appear on a given exam in advance. The free exam sandbox at aka.ms/examdemo lets you interact with every question type in the real interface.
  • Some questions on every Microsoft exam are unscored trial items used to gather psychometric data. They are not marked, so answer every question.
  • Non-native English speakers taking the exam in English can request an additional 30 minutes.

How scoring works

Score scale

1 to 1,000

Score needed to pass

700

Roughly what that means raw

Not published. Microsoft states directly that 700 is a scaled score and may not equal 70 percent of the points, and that the section bar chart cannot be used to calculate how many questions you answered correctly.

When results arrive

Within minutes of finishing for exams without labs; the score also appears in your Microsoft Learn profile within 24 hours

How the scale is built

Microsoft awards one point for each correctly answered component, so multi-part questions can earn all, some, or none of their points. Nothing is deducted for a wrong answer. Raw points are converted to a scaled score so that forms of differing difficulty are equated: an easier set of questions requires more raw points to reach 700, a harder set requires fewer.

  • The score report gives one overall numeric score, a pass or fail status, and a bar chart of performance per skill area. It does not give a numeric score per section.
  • Scoring zero in one skill area is possible and not an error. Some areas carry only a few questions.
  • Unscored trial questions are mixed in and are not identified, so treat every item as if it counts.
  • Microsoft will not disclose which questions you answered incorrectly, on either a pass or a fail.
  • Identical scores across repeat attempts are normal and Microsoft treats them as evidence of exam reliability rather than a fault.

Pacing and time budget

42-46 questions in 45 minutes gives you About 60 seconds at the top of the published range, about 64 seconds at the bottom per question.

At this pointYou should have answered
10 min10
20 min20
30 min30
40 min42
  • Five of the 45 minutes are the built-in break allowance. If you do not take a break, that time is available for questions.
  • Drag and drop, build list, and hot area items take longer than plain multiple choice, so bank time on the concept questions in the 40-45 percent group.
  • A question you cannot place within 90 seconds is costing you two later questions. Pick the best option, mark it for review, and move on.
  • Come back to marked questions before any break, because the break closes access to everything you have already seen.
  • The exam page and score report do not tell you how many of your items were unscored trial questions, so do not skip anything on the theory that it might not count.

Where the marks are

TopicWeightWhy it scores
Deploying a model in the Foundry portal and interacting with itWithin the 55-60% Foundry implementation groupIt is the first listed task in the largest weighted group and it is the prerequisite skill for the SDK, agent, speech, vision, and extraction tasks. Every other Foundry item assumes you can get a model deployed.
Writing system and user promptsWithin the 55-60% Foundry implementation groupPrompt construction appears as its own bullet and also underpins the multimodal speech and vision tasks, where the prompt is how audio or an image enters the model.
Azure AI Content Understanding across documents, images, audio, and videoWithin the 55-60% Foundry implementation groupIt is the named service in all four bullets of the information-extraction sub-skill. It also carries the highest chance of catching candidates out, because no AI-900 material covers it.
Creating and testing a single-agent solutionWithin the 55-60% Foundry implementation groupAgents are new to this exam relative to AI-900 and occupy two of the five bullets in the generative AI sub-skill, one for the portal build and one for the client application.
The six responsible AI considerationsWithin the 40-45% concepts groupSix named bullets means six identifiable question hooks, and the answers are pure recall once you know the list. This is the cheapest block of marks on the exam.
Matching a workload to generative, agentic, text, speech, vision, or extractionWithin the 40-45% concepts groupMicrosoft names the categories explicitly in the study guide, and scenario questions test the mapping directly. Getting the category right usually eliminates two of the four options immediately.
Choosing a model and its deployment configurationWithin the 40-45% concepts groupThree bullets cover how generative models work, selecting a model by capability, and picking deployment options and configuration parameters. The same knowledge is reused in the Foundry implementation questions.

The numbers

$99 USD in the United States, priced by the country or region of proctoring

Exam fee

Source: Microsoft Learn, Exam AI-901 details page

700 on a 1 to 1,000 scale

Passing score

Source: Microsoft Learn, Exam scoring and score reports

45 minutes exam duration, 65 minutes seat time

Exam duration and seat time

Source: Microsoft Learn, Exam duration and exam experience

42 to 46 questions

Question count

Source: Certiport (Pearson VUE), Exam lengths educator resource

24 hours after the first failure, 14 days between all later attempts, maximum five attempts per 12 months

Retake waiting periods

Source: Microsoft Learn, Exam and Applied Skills assessment lab retake policies

Not published

Pass rate

Source: Microsoft Learn publishes no pass-rate figures on its exam, scoring, or policy pages

If you fail

Wait before retaking

24 hours after a first failure, then 14 days between every subsequent attempt

Attempt limit

Five attempts in any 12-month period, counted from the date of the first attempt. After a fifth failure you become eligible again 12 months after that first attempt.

Retake fee

Full exam fee for every attempt. The Exam Replay offer bundles a first booking with a second attempt at less than two full fees and must be bought before the first sitting.

A waiver of the waiting period is granted only for internet connectivity issues or equipment failure during the exam, and the request must include a Pearson VUE case number. Waivers are not granted for any other reason. You cannot retake an exam you have already passed unless the certification has expired, which does not apply here because Fundamentals certifications do not expire.

Exam Domains & Topics

Describe principles of responsible AI
Not published at sub-skill level; part of the Identify AI concepts and capabilities group, which Microsoft weights at 40-45%

Microsoft lists six named responsible AI considerations and the exam asks you to recognise which one a scenario violates. The six are fairness, reliability and safety, privacy and security, inclusiveness, transparency, and accountability. Questions are usually short scenarios where a system behaves badly and you pick the principle at stake.

Key Topics to Master:

  • Considerations for fairness in an AI solution
  • Considerations for reliability and safety in an AI solution
  • Considerations for privacy and security in an AI solution
  • Considerations for inclusiveness in an AI solution
  • Considerations for transparency in an AI solution
  • Considerations for accountability in an AI solution
Identify AI model components and configurations
Not published at sub-skill level; part of the Identify AI concepts and capabilities group, which Microsoft weights at 40-45%

This sub-skill covers how generative AI models work at a conceptual level and how you choose among them. Microsoft asks candidates to match a model to a stated capability requirement and to pick deployment options and configuration parameters that fit a workload. Expect questions on why one model family suits a task that another does not.

Key Topics to Master:

  • How generative AI models work
  • Selecting an appropriate AI model based on capabilities
  • Model deployment options in Microsoft Foundry
  • Configuration parameters such as temperature and token limits
  • Trade-offs between model size, cost, and latency
Identify AI workloads
Not published at sub-skill level; part of the Identify AI concepts and capabilities group, which Microsoft weights at 40-45%

Microsoft names the workload categories explicitly: generative and agentic AI, text analysis, speech, computer vision, and information extraction. You need to place a described business problem in the right category and name the technique that solves it. The text analysis techniques listed are keyword extraction, entity detection, sentiment analysis, and summarisation.

Key Topics to Master:

  • Scenarios for generative and agentic AI workloads
  • Keyword extraction, entity detection, sentiment analysis, summarisation
  • Speech recognition and speech synthesis capabilities
  • Computer vision and image-generation model capabilities
  • Techniques to extract information from text, images, audio, and video
Implement generative AI apps and agents by using Foundry
Not published at sub-skill level; part of the Implement AI solutions by using Microsoft Foundry group, which Microsoft weights at 55-60%

The largest hands-on block. Microsoft expects you to write effective system and user prompts, deploy a model and interact with it in the Foundry portal, build a lightweight chat client with the Foundry SDK, and create and test a single-agent solution in the portal plus a lightweight client for that agent. Python syntax appears in the answer options.

Key Topics to Master:

  • Writing system prompts and user prompts
  • Deploying a model in the Microsoft Foundry portal
  • Building a chat client application with the Foundry SDK
  • Creating and testing a single-agent solution in the Foundry portal
  • Building a lightweight client application for an agent
Implement AI solutions for text and speech by using Foundry
Not published at sub-skill level; part of the Implement AI solutions by using Microsoft Foundry group, which Microsoft weights at 55-60%

Covers building a lightweight application that performs text analysis, responding to spoken prompts using a deployed multimodal model, and building an application with Azure Speech in Foundry Tools. The distinction between calling a multimodal model directly and calling the dedicated Azure Speech service is tested, because both can handle audio input.

Key Topics to Master:

  • Building a lightweight text analysis application
  • Responding to spoken prompts with a deployed multimodal model
  • Azure Speech in Foundry Tools
  • Speech to text and text to speech configuration
  • Choosing between multimodal models and dedicated speech services
Implement AI solutions with computer vision and image-generation capabilities by using Foundry
Not published at sub-skill level; part of the Implement AI solutions by using Microsoft Foundry group, which Microsoft weights at 55-60%

Three tasks are listed: interpret visual input inside a prompt using a deployed multimodal model, create new visual outputs with generative models, and build a lightweight application that includes vision capabilities. Image understanding and image generation are treated as separate capabilities with separate model choices, and the exam tests that separation.

Key Topics to Master:

  • Passing images into prompts for a multimodal model
  • Generating images with generative models
  • Building a lightweight vision-enabled application
  • Choosing between image understanding and image generation models
  • Handling image input formats and size limits
Implement AI solutions for information extraction by using Foundry
Not published at sub-skill level; part of the Implement AI solutions by using Microsoft Foundry group, which Microsoft weights at 55-60%

Azure AI Content Understanding is the named service across all four listed tasks: extracting information from documents and forms, from images, from audio, and from video, plus building a lightweight application with information extraction capabilities. This service did not appear on AI-900, so older study material will not cover it at all.

Key Topics to Master:

  • Azure AI Content Understanding in Foundry Tools
  • Extracting fields from documents and forms
  • Extracting information from images
  • Extracting information from audio and video
  • Building an application with information extraction capabilities

Recommended Study Plan

Week 1: Confirm the right exam and set up the environment
6
  • 1Read the AI-901 exam page and confirm the retirement date field reads none
  • 2Download the AI-901 study guide and copy the two weightings, 40-45 percent and 55-60 percent, into your notes
  • 3Create an Azure account and a Microsoft Foundry project so you have somewhere to deploy models
  • 4Register a Microsoft Learn profile with a personal Microsoft account rather than a work or school account
  • 5Run the exam sandbox at aka.ms/examdemo and try every question type once
Week 2: Responsible AI and generative model concepts
7
  • 1Memorise the six responsible AI considerations in Microsoft's order and write one failure example for each
  • 2Read Microsoft's documentation on how generative AI models work
  • 3Build a table comparing model families by capability, context length, and modality
  • 4Practise picking a model from a stated capability requirement, five scenarios minimum
  • 5Note which configuration parameters change output randomness versus output length
Week 3: AI workload identification
7
  • 1Write definitions for generative AI, agentic AI, text analysis, speech, computer vision, and information extraction
  • 2Practise keyword extraction, entity detection, sentiment analysis, and summarisation on the same sample paragraph
  • 3List what speech recognition does that speech synthesis does not, and the reverse
  • 4Compare computer vision tasks against image-generation tasks in one table
  • 5Draft ten one-line scenarios and label the correct workload category for each
Week 4: Foundry portal basics and model deployment
8
  • 1Deploy one model in the Microsoft Foundry portal and record every field you had to set
  • 2Interact with the deployed model in the portal chat playground
  • 3Change the temperature and token settings and note how the output shifts
  • 4Read the Foundry deployment options documentation and note the differences between them
  • 5Delete and redeploy the model so the workflow is automatic
Week 5: Prompts, the Foundry SDK, and a chat client
9
  • 1Write five system prompts that constrain tone, format, and refusal behaviour
  • 2Install the Foundry SDK and authenticate against your project
  • 3Build a chat client of under 50 lines of Python that sends a user message and prints the reply
  • 4Add a system prompt to that client and confirm the behaviour change
  • 5Read the SDK reference for the chat completion call and learn the parameter names
Week 6: Agents
9
  • 1Create a single-agent solution in the Foundry portal and give it one tool
  • 2Test the agent in the portal and read its run trace
  • 3Build a lightweight client application that calls the agent
  • 4Write down the difference between calling a model directly and calling an agent
  • 5Practise describing when an agent is the right choice and when a plain prompt is enough
Week 7: Text, speech, vision, and Content Understanding
10
  • 1Build a lightweight application that runs text analysis on a sample document
  • 2Send a spoken prompt to a deployed multimodal model and capture the reply
  • 3Use Azure Speech in Foundry Tools for both speech to text and text to speech
  • 4Pass an image into a prompt and then generate a new image with a generative model
  • 5Run Azure AI Content Understanding over one document, one image, and one audio file
Week 8: Practice assessment and exam conditions
8
  • 1Take the practice assessment on AI Skills Navigator and record which of the seven sub-skill areas scored lowest
  • 2Re-run the weakest two sub-skill areas as hands-on labs rather than reading
  • 3Do one full 45-minute timed run with no notes and no browser tabs open
  • 4Run the Pearson VUE system test on the exact machine and connection you will use if testing online
  • 5Check that the name on your Microsoft Certification profile matches your government-issued ID exactly

Ready to pass Azure AI Fundamentals (AI-901)?

Get 500+ practice questions, video walkthroughs, and a pass guarantee.

94% pass rate on first attempt
$50$25

Best Study Resources

AI-901 study guide

Official study guide

Microsoft's own skills-measured outline, listing both weighted groups and every sub-skill bullet the exam draws from. It is the only document that tells you the 40-45 and 55-60 percent split.

Free

Exam AI-901 details page

Official exam page

Lists the $99 USD price, the 13 exam languages, the 700 passing score, the retirement date field, and the scheduling links for Pearson VUE and Certiport.

Free

AI Skills Navigator practice assessment

Practice test

Microsoft moved the AI-901 practice assessment to AI Skills Navigator. You must sign in to launch it. It reports performance by skill area, which maps to the study guide headings.

Free

Microsoft exam sandbox

Interface demo

A working copy of the exam user interface with every question type, the candidate agreement, the timer, the mark-for-review control, and the comment box. Use it once before booking.

Free

Course AI-901T00: Introduction to AI in Azure

Instructor-led course

Microsoft's official instructor-led course for this exam. It mixes AI concepts with the Azure technology skills used to build AI solutions, aimed at people starting a career in AI development.

Priced by the delivering Microsoft partner

Microsoft Foundry documentation

Product documentation

The reference for model deployment, the Foundry portal, agent creation, and the Foundry SDK. Because 55 to 60 percent of the exam is Foundry implementation, this is the highest-return reading on the list.

Free

Azure AI Content Understanding documentation

Product documentation

Covers extraction from documents, forms, images, audio, and video. This service is named in all four information-extraction tasks in the study guide and did not exist on the AI-900 outline.

Free

Exam Replay

Retake offer

A Microsoft offer that bundles an exam booking with a second attempt. Buying it before the first attempt is cheaper than paying two full fees if you expect a close result.

Priced above the standard exam fee

Microsoft exam duration and exam experience page

Policy page

Source for the 45-minute exam duration and 65-minute seat time for Fundamentals exams, the five built-in break minutes, and the rule that Microsoft Learn access is unavailable on Fundamentals exams.

Free

Microsoft exam retake policy

Policy page

States the 24-hour wait after a first failure, the 14-day wait between all later attempts, and the limit of five attempts in a 12-month period counted from the first attempt.

Free

Common Mistakes to Avoid

Studying AI-900 material for AI-901.

AI-900 retired on 30 June 2026 and its outline had five skill areas built around Azure AI services. AI-901 has two groups built around Microsoft Foundry. Any course, book, or practice set that lists computer vision workloads and NLP workloads as separate weighted domains is the old exam.

Treating the exam as pure theory and never opening the Foundry portal.

The Implement AI solutions by using Microsoft Foundry group is 55 to 60 percent of the scored content and every one of its bullets starts with a verb like create, deploy, build, or extract. Deploy a model, build a chat client, and create an agent at least once each.

Skipping Python because the exam is labelled fundamentals.

Microsoft's own audience profile says you need knowledge of Python coding syntax and programming techniques. You will not write full programs, but you will read short snippets and pick the correct SDK call or parameter.

Learning five responsible AI principles instead of six.

Microsoft lists six: fairness, reliability and safety, privacy and security, inclusiveness, transparency, and accountability. Reliability and safety is one item, and privacy and security is one item. Splitting or merging them loses marks on scenario questions.

Ignoring Azure AI Content Understanding.

It is the named service in all four information-extraction tasks: documents and forms, images, audio, and video. Run it against one file of each type so you know what the output looks like.

Taking a break mid-exam without answering visible questions first.

Microsoft warns that once a break starts you cannot return to any question you have already seen, even unanswered ones or those marked for review. Answer everything on screen before selecting Take a break.

Booking with a work or school account.

Microsoft strongly recommends registering with a personal Microsoft account. Exam records tied to an organisational account are lost and unrecoverable if you leave that organisation.

Expecting Microsoft Learn access during the exam.

The in-exam Microsoft Learn browser is available on associate and expert role-based exams only. Fundamentals and GitHub exams do not have it, so there is nothing to look up.

Assuming a score of 700 means 70 percent of questions correct.

Microsoft reports a scaled score on a 1 to 1,000 range and states directly that it may not equal 70 percent of the points. The cut is set against question difficulty, so an easier form needs more points and a harder form needs fewer.

Having a name mismatch between the ID and the certification profile.

Microsoft requires the name on your government-issued ID to match your Microsoft Certification profile exactly. Legal name-change forms and ID renewal forms are not accepted for online proctored exams, so fix the profile weeks ahead.

Exam Day Tips

  • 1

    Budget roughly one minute per question. With 45 minutes and a published range of 42 to 46 questions, there is no slack for a long stall.

  • 2

    Do not plan to use the five built-in break minutes unless you need them. The exam clock keeps running during a break and you lose access to every question you have already seen.

  • 3

    For online testing, check in from 30 minutes before your appointment time. Check-in closes 15 minutes after the appointment time.

  • 4

    Run the Pearson VUE system test again on exam day, on the same machine, network, and physical location you will test from.

  • 5

    Clear pens, notepads, sticky notes, printed pages, and books out of arm's reach before the room scan, and unplug or turn away extra monitors.

  • 6

    Keep your phone within arm's reach only until check-in ends, then place it out of reach but close enough to hear if the proctor calls.

  • 7

    A beverage is allowed without an accommodation as long as it is in a transparent, spill-proof container. Food, gum, and smoking need an approved accommodation.

  • 8

    Mark a question for review only if you can afford to come back before your next break. There is no post-break return.

  • 9

    Answer every question. Microsoft applies no penalty for guessing, and unanswered items simply score zero.

  • 10

    Expect your score and a bar chart by skill area within minutes of finishing, and expect it in your Learn profile within 24 hours.

Career Paths & Salary Ranges

Computer support specialist handling AI tooling

Front-line support staff increasingly field questions about copilots, chat assistants, and model behaviour. AI-901 gives the vocabulary to triage those tickets and to explain why a model produced a given answer.

$61,550 US median (BLS, May 2024)

Web developer or digital designer building AI-assisted interfaces

The Foundry SDK chat client task in the AI-901 outline is the same pattern used to bolt a model onto a web front end. The certification confirms you can deploy a model and call it rather than only consume a third-party API.

$95,380 US median (BLS, May 2024)

Computer systems analyst scoping AI projects

Analysts translate business problems into workload categories. The exam's workload identification group covers exactly that: matching a described problem to generative AI, text analysis, speech, computer vision, or information extraction.

$103,790 US median (BLS, May 2024)

Database administrator or architect feeding AI systems

Content Understanding pipelines that extract fields from documents and audio land their output in a data store. Knowing what the extraction service emits helps design the schema that receives it.

$123,100 US median (BLS, May 2024)

Software developer building agent applications

AI-901 requires creating a single-agent solution in the Foundry portal and a client application for it. That is the entry point to agent development work, and it is the natural bridge to Microsoft's higher-tier AI engineering credentials.

$131,450 US median (BLS, May 2024)

Prerequisites & Requirements

  • No formal prerequisite exam, degree, or work experience is required to book AI-901.
  • Microsoft's audience profile expects conceptual knowledge of AI solutions in Azure and the foundational technical skills to work with them.
  • Knowledge of Python coding syntax and programming techniques is listed by Microsoft as a requirement for this exam.
  • Familiarity with Azure resources is expected, including core cloud concepts such as cloud storage, cloud compute, and cloud-based authentication and authorisation.
  • Familiarity with REST APIs, SDKs, and CLIs is stated on the exam page.
  • A Microsoft Learn profile registered with a personal Microsoft account, plus a government-issued ID whose name matches that profile exactly.
  • An Azure subscription is not required to sit the exam but is needed to complete the hands-on Foundry work the exam assumes.

Frequently Asked Questions

Is AI-900 or AI-901 the current Azure AI Fundamentals exam?

AI-901 is the current exam. Microsoft retired AI-900 on 30 June 2026 and the AI-901 exam page lists no retirement date. Both exams lead to the same Microsoft Certified: Azure AI Fundamentals certification, so anyone who passed AI-900 before it retired already holds the credential and does not need to sit AI-901.

What does AI-901 cost?

The exam is $99 USD in the United States. Microsoft states that the price is based on the country or region in which the exam is proctored, so the local figure differs elsewhere and taxes are not included. Microsoft also notes that pricing does not reflect promotional offers or the reduced pricing available to Microsoft Certified Trainers and Microsoft Partner Network members.

How long do I have to wait to retake AI-901 if I fail?

You must wait 24 hours after a first failure before retaking. A 14-day waiting period applies between all subsequent attempts. You may not take the exam more than five times within a 12-month period counted from your first attempt, and if you fail five times you become eligible again 12 months after that first attempt date.

Do I pay again for a retake?

Yes. Microsoft's retake policy states that you must pay to retake the exam. The Exam Replay offer bundles a booking with a second attempt for less than two separate full fees, and it has to be purchased before the first attempt. A waiver of the waiting period is granted only where a Pearson VUE case number confirms an internet connectivity issue or equipment failure during the exam.

What happens if I fail?

You receive a numeric score, a fail status, and a bar chart showing performance in each skill area. Microsoft does not tell you which specific questions you answered incorrectly. Use the bar chart to find the weakest area and note that shorter bars mean weakness, but the chart cannot be converted back into a count of correct answers.

Does the Azure AI Fundamentals certification expire?

No. Microsoft states that Fundamentals certifications do not expire. Associate, expert, and specialty certifications expire annually and are renewed by passing a free unproctored online assessment on Microsoft Learn within a six-month eligibility window. That renewal process does not apply to AI-901.

What is the passing score and how is it calculated?

The passing score is 700 on a scale of 1 to 1,000. Microsoft states this is a scaled score that may not equal 70 percent of the points, because the cut is set against the knowledge and skills needed to demonstrate competence and the difficulty of the questions asked. Easier question sets need more points to pass and harder sets need fewer.

When do I get my result?

For exams without labs you receive results within minutes of finishing, along with a report showing your score and performance feedback. The score appears in your Learn profile within 24 hours. Exams taken through Certiport show final score and pass or fail status on screen immediately, with results in the Learn profile or Certiport dashboard within 24 hours.

What ID do I need?

You must present a current government-issued ID whose name matches the name on your Microsoft Certification profile exactly. Legal name-change forms and ID renewal forms are not accepted for online proctored exams. If you are mid-renewal on your identification, book a test-centre appointment or wait until the new ID arrives.

How does online proctoring work?

Pearson VUE delivers online exams through OnVUE, with a proctor monitoring by webcam and microphone. You check in from 30 minutes before your appointment and no later than 15 minutes after it. You photograph your face, your ID, and your workspace with a mobile phone, a greeter reviews them, and then you enter the proctor queue. Launch and check-in takes about 15 minutes.

What can I have in the room during an online exam?

Almost nothing. Books, notepads, sticky notes, typed notes, pens, markers, pencils, and whiteboards must be out of arm's reach. Extra monitors must be unplugged and turned away and other computers switched off. No hand-held devices, watches, wallets, purses, hats, bags, or coats. A beverage in a transparent spill-proof container is permitted without an accommodation.

Is scratch paper or a whiteboard allowed?

No. Microsoft's testing environment requirements list notepads, sticky notes, typed notes, and writing instruments including whiteboards among the items that must be kept out of arm's reach during online proctored exams. Plan to work entirely on screen.

How do I request accommodations?

Request accommodations through Microsoft's request-accommodations process before you schedule the exam, so the arrangement is in place at booking. Assistive devices need prior approval because the secure browser blocks third-party applications during the exam. Separately, non-native English speakers taking the exam in English can request an additional 30 minutes.

How many questions are on AI-901?

Certiport's published exam length list gives AI-901 a range of 42 to 46 questions. Microsoft does not publish a fixed count for this exam and states that most of its certification exams contain between 40 and 60 questions, with the number subject to change as the exam is updated. Some of the questions you see are unscored trial items.

How long is the exam?

Microsoft publishes 45 minutes of exam duration and 65 minutes of seat time for all Fundamentals exams. Exam duration is the working time; seat time also covers reviewing instructions, accepting the candidate agreement, and leaving comments afterwards. Five minutes of break time is built into the 45 minutes, and Microsoft removed questions to make room for it.

Can I use Microsoft Learn during the exam?

No. The in-exam Microsoft Learn browser is available on associate and expert role-based exams only. Microsoft states explicitly that access to Learn is not available on Fundamentals exams or GitHub exams. Nothing is added to the clock for lookups because there are none.

Can I reschedule or cancel?

Yes, but you must do it at least 24 hours before the scheduled start or the fee is forfeited. A company-purchased voucher is forfeited on the same terms. Reschedule and cancel through your Microsoft Learn profile, which redirects you to the delivery provider. A no-show may result in your company being charged a no-show fee.

How does AI-901 compare with AI-102?

AI-901 is the fundamentals-tier exam at $99 with 45 minutes of exam time and no prerequisite experience. AI-102, the Azure AI Engineer Associate exam, is an associate role-based exam, which Microsoft schedules at 100 or 120 minutes depending on whether labs are present, and which carries an annual free renewal requirement that AI-901 does not. Take AI-901 first if you have not deployed a model in Microsoft Foundry before.

Does Microsoft publish a pass rate for AI-901?

No. Microsoft does not publish pass rates for any certification exam. Its scoring documentation covers the 1 to 1,000 scale, the 700 cut, and how score reports are read, but contains no candidate volume or pass percentage data. Any pass rate figure you see for AI-901 comes from a third party rather than from Microsoft.

Do I need an Azure subscription to prepare?

Not to sit the exam, but you need one to do the work the exam assumes. Six of the study guide's tasks require deploying a model in the Foundry portal, calling it from the Foundry SDK, creating an agent, and running Azure AI Content Understanding. Reading about those steps is measurably weaker preparation than performing them once each.

50% OFF

Pass Azure AI Fundamentals (AI-901) — Guaranteed

94% pass rate on first attempt

500+ Real QuestionsUpdated weekly
Video Walkthroughs20+ hours
Pass or Full RefundGuaranteed
Lifetime AccessFree updates
SAVE $25
$25
$5050% OFF

One-time • Lifetime access

Secure Instant
4.9/5 (2,847 reviews)
30-Day Guarantee — Pass or get 100% refund