AI-103 practice test

AI-103 Practice Test, Practice Questions and Exam Prep From Your Own Azure AI Notes

AI-103 replaced AI-102, and Microsoft has not released an official practice assessment for it. Upload your Foundry notes, Azure AI documentation or prep PDF, and the AI writes unlimited AI-103 practice questions with an answer key in seconds.

Your study files are processed securely and deleted automatically after your practice questions are built.

Upload your Azure AI notes or prep PDF and generate your first question set

Click to upload or drag and drop

PDF, DOCX, PPTX, TXT, JPG, JPEG, PNG, HEIC, ODP, ODT, BMP, or TIFF

up to 20MB

Please wait, your quiz is being created...

Uploading...

AI-103 is the exam that replaced AI-102, and AI-102 is already gone. Its title is Developing AI Apps and Agents on Azure, and passing it earns the Azure AI Apps and Agents Developer Associate certification. AI-102 and the Azure AI Engineer Associate retired on June 30, 2026: Microsoft's page for it now reads "This certification and the renewal assessment are retired." AI-103 needs 700 out of 1000 to pass, has five domains, and its biggest slice is generative AI and agentic solutions at 30 to 35 percent. The detail that catches people out: the phrase "Azure OpenAI" appears zero times in the AI-103 outline. It appeared five times in AI-102's. Everything now runs through Microsoft Foundry.

Last updated July 2026

AI-102 is retired, and a lot of what is still being sold does not know it

This is not a scheduled change you can plan around. It already happened. Go to Microsoft's Azure AI Engineer Associate page today and the first thing on it is a warning block: "This certification and the renewal assessment are retired." AI-102 stopped being bookable on June 30, 2026. Yet search for AI-102 prep and you will still find courses, bootcamps and question banks selling preparation for it as though nothing has changed.

AI-102 was not alone. It was part of a wider rebrand in which Microsoft retired six certifications and replaced them with Cloud-and-AI successors. Three of those retirement dates have now passed.

Retiring certification (exam) Retirement date Status today Replacement
Azure Data Scientist Associate (DP-100)June 1, 2026RetiredAI-300
Azure AI Engineer Associate (AI-102)June 30, 2026RetiredAI-103, this exam
Azure AI Fundamentals (AI-900)June 30, 2026RetiredAI-901
Azure Developer Associate (AZ-204)July 31, 2026RetiringAI-200
Azure Security Engineer Associate (AZ-500)August 31, 2026RetiringSC-500
Windows Server Hybrid Admin (AZ-800, AZ-801)September 30, 2026RetiringAZ-802

If you already passed AI-102, your certification stays valid until its existing expiration date. What you have lost is the renewal. The renewal assessment retired with the exam, so when your credential lapses there is no free online assessment waiting to extend it. The way to stay current as an Azure AI developer is AI-103.

Passing score
700 / 1000
Biggest domain
30 to 35%
"Azure OpenAI" in the outline
0 times
Practice questions
Unlimited

Microsoft deleted the Azure OpenAI vocabulary from this exam

Here is a ten second test you can run yourself. Open Microsoft's AI-103 study guide, press Ctrl+F, and search for "Azure OpenAI." You will get nothing. Now do the same on the archived AI-102 guide and you get five hits. We counted every load-bearing term in both official outlines. This is what changed.

Term In AI-102's outline (retired) In AI-103's outline (current) What it means for your prep
Azure OpenAI50The service names are gone. Models are deployed and consumed through Foundry now.
Agent1121Nearly doubled. Agents are the center of gravity of this exam.
Foundry2313Still the platform, and now the only vocabulary for model work.
Prompt56Prompt engineering survives, but as a tuning topic, not a headline.
Retrieval, RAG05RAG is now named explicitly as a build task.
Responsible AI43Still examined, now framed as guardrails and agent governance.

The trap is that the underlying work barely moved. You are still deploying models and grounding them against your data. What moved is what Microsoft calls it, and an exam tests the words on its own outline. A question that used to say "configure an Azure OpenAI deployment" now says "configure model and agent deployments in Foundry." If you revise from an AI-102 course you will know the concept and miss the term.

This is exactly the kind of drift a recycled question bank cannot fix, because the bank was written against the old vocabulary. Feeding the current Microsoft documentation into a generator and letting it build a question bank from your own Foundry notes keeps your practice in the same language as the exam.

AI-103 skills measured, straight from Microsoft's study guide

Five domains, current as of the April 16, 2026 update. Two of them carry 55 to 65 percent of the exam between them, and both are about generative AI and agents.

Domain What is actually in it Weight
Implement generative AI and agentic solutionsDeploying and consuming LLMs, small models, code models and multimodal models. Implementing retrieval-augmented generation in an application. Designing tool-augmented flows and multistep reasoning pipelines. Evaluating apps for fabrications, relevance, quality and safety. Building agents: defining roles, goals, conversation tracking and tool schemas, function calling, conversation memory, orchestrated multi-agent solutions, and autonomous workflows with approval controls. Tuning generation behavior, including prompt engineering. The largest domain.30 to 35%
Plan and manage an Azure AI solutionChoosing the right model for a task across LLMs, small language models and multimodal models. Choosing Foundry services for generation, grounding, vector search, agent workflows. Choosing retrieval and indexing methods. Designing infrastructure and deployment options, and wiring Foundry projects into CI/CD pipelines. Managing quotas, scaling, rate limits and cost footprints. Monitoring drift, safety events and grounding quality. Security via managed identity, private networking and keyless credentials. Responsible AI: safety filters, guardrails, evaluators, trace logging, provenance metadata, and governing agent behavior with oversight modes.25 to 30%
Implement computer vision solutionsVision capability built through Foundry rather than standalone Cognitive Services endpoints, including multimodal interpretation of visual input and generative image output.10 to 15%
Implement text analysis solutionsKeyword extraction, entity detection, sentiment analysis and summarization, plus speech, delivered through Foundry Tools.10 to 15%
Implement information extraction solutionsExtracting structured information from documents, forms, images, audio and video using Azure Content Understanding. This is the successor to a lot of what used to be Document Intelligence and Form Recognizer work.10 to 15%

Notice the shape of it. The three 10 to 15 percent domains, vision, text and extraction, are the classic Cognitive Services surface that used to be most of AI-102. They now add up to a minority of the exam. The majority is generative AI, agents, and the governance around them.

Microsoft does not publish a question count for AI-103. Its only general guidance is that most exams "typically contain between 40-60 questions," and that the number can vary. It publishes no pass rate for any exam. As an associate role-based exam AI-103 is open book: you can browse learn.microsoft.com while you sit it, excluding Q&A, practice assessments and your profile, with no extra time granted. Microsoft's exam pricing feed does not yet list a US price for AI-103. Associate role-based exams are normally $165 in the United States, but confirm the figure at checkout rather than trusting any site that prints it as fact.

AI-103 or AI-200? The two exams people mix up constantly

Microsoft launched two new AI-coded developer exams in the same wave, and their names do almost nothing to tell them apart. People pick the wrong one. The single cleanest way to separate them is to count one word in each official outline.

AI-103 AI-200
Exam titleDeveloping AI Apps and Agents on AzureDeveloping AI Cloud Solutions on Azure
"Agent" in the outline21 times0 times
"Foundry" in the outline13 times0 times
"Container" in the outline0 times11 times
Certification earnedAzure AI Apps and Agents Developer AssociateAzure AI Cloud Developer Associate
ReplacesAI-102 (retired June 30, 2026)AZ-204 (retires July 31, 2026)
Take it if youBuild with models: RAG, agents, prompts, evaluation, content safetyBuild the platform under them: containers, vector stores, messaging, tracing

Read the two bottom rows together and the choice is usually obvious within ten seconds. If your day job is writing the code that calls a model, grounds it, evaluates it and stops it saying something dangerous, that is AI-103. If your day job is Cosmos DB, Container Apps, Service Bus and OpenTelemetry, and your AI exposure is storing embeddings and serving them fast, that is AI-200, and the AZ-204 skills you already have transfer directly.

The one thing you should not do is assume the higher number is the more advanced exam. They are peers, aimed at different jobs. AI-200 is not AI-103 with more content, and AI-103 is not a prerequisite for anything.

There is no official AI-103 question bank yet

Microsoft states it plainly on the certification page: the practice assessment is "not currently available," and practice assessments are "usually available within 8 weeks of the exam being out of beta and generally available." So for AI-103 right now there is no official bank, from anyone, at any price.

That has a specific consequence worth being blunt about. Every product currently advertising "real AI-103 exam questions" or a precise question count was written without access to an official source. Microsoft has never published a per-exam question count and has never published a pass rate, for AI-103 or for anything else. When a site prints "AI-103: 65 questions, 78% pass rate," both numbers were made up, and the confidence is the product.

A newer exam also means the recycled banks that do exist are recycled from AI-102, which is the one exam whose vocabulary Microsoft just rewrote. You would be drilling the wrong words.

The alternative that actually works for a new exam is to generate questions from the source material the exam is built on. Microsoft publishes the outline and the documentation behind every bullet in it. That material is current by definition.

How to build AI-103 practice questions when no official bank exists

Generate from the current source rather than waiting for a bank that will arrive late and may still be written in the old vocabulary.

1
Upload the current source
Microsoft's Foundry documentation, the AI-103 study guide itself, or your own team's architecture notes.
2
Weight the agent domain
Generative AI and agents is 30 to 35 percent, the biggest slice, and the newest material for most AI-102 holders.
3
AI writes the questions
Exam-style items with an answer key and explanations, in the vocabulary the current outline actually uses.
4
Drop the old service names
If your notes still say Azure OpenAI resource, they predate this exam. Regenerate from Foundry material.

AI-103 questions, answered

What is Exam AI-103?
AI-103 is Microsoft's exam titled Developing AI Apps and Agents on Azure. Passing it earns the Microsoft Certified: Azure AI Apps and Agents Developer Associate certification. It is the replacement for AI-102, the Azure AI Engineer Associate, whose certification page now states that the certification and its renewal assessment are retired.
Is AI-102 retired?
Yes. AI-102 and the Azure AI Engineer Associate certification retired on June 30, 2026. Microsoft's certification page now carries the warning that this certification and the renewal assessment are retired. If you already hold it, it stays valid until it expires, but you can no longer take or renew it. AI-103 is the path forward.
Does AI-103 still cover Azure OpenAI?
The phrase Azure OpenAI appears zero times in Microsoft's AI-103 skills outline. It appeared five times in AI-102's. The work did not disappear, the vocabulary changed: models are now deployed and consumed through Microsoft Foundry, which the outline names 13 times. Study Foundry, not the Azure OpenAI resource names.
What are the AI-103 domain weights?
Five domains. Implement generative AI and agentic solutions is the largest at 30 to 35 percent. Plan and manage an Azure AI solution is 25 to 30 percent. Computer vision, text analysis, and information extraction are each 10 to 15 percent.
What is the difference between AI-103 and AI-200?
AI-103 is the agents exam and AI-200 is not. The word agent appears 21 times in AI-103's outline and zero times in AI-200's. AI-200 is a backend exam about containers, vector databases and observability, and it replaces AZ-204. AI-103 replaces AI-102 and covers Foundry, RAG, multi-agent orchestration and responsible AI.
Is there an official AI-103 practice assessment?
No. Microsoft's certification page states the practice assessment is not currently available, and that practice assessments usually appear within 8 weeks of an exam leaving beta and becoming generally available. There is no official AI-103 question bank, so any product claiming real AI-103 exam questions was not written from one.
What score do you need to pass AI-103?
700 or greater on a scale of 1 to 1000. Microsoft notes that because this is a scaled score it may not equal 70 percent of the points. Microsoft publishes no question count and no pass rate for AI-103 or for any exam, so treat any exact figure you see as invented.
Does my AI-102 certification still count?
Yes, until it expires. Microsoft's policy is that certifications already earned remain valid through their existing expiration date even after the exam retires. What you lose is the ability to renew it, because the renewal assessment retired with the exam. To stay current as an Azure AI developer you move to AI-103.

PDFQuiz is not affiliated with, endorsed by, or sponsored by Microsoft. Microsoft, Azure and Microsoft Foundry are trademarks of the Microsoft group of companies. This generator builds practice questions from material you upload and is a study aid, not a substitute for Microsoft's official learning paths. AI-103 is a new exam and its details are still settling: always confirm on Microsoft's own pages before you book.

Related study tools

The exam AI-103 is most often confused with is AI-200, which replaces the Azure developer exam and has no agents on it at all. If you are coming up through the fundamentals, AI-901 is the exam that replaced AI-900, and the AZ-900 practice test generator covers the cloud fundamentals exam that is not retiring. On the machine learning side of the same wave, AI-300 replaces DP-100. The exam most Azure AI developers hold alongside this one is AZ-104, and for solution design there is AZ-305. For any source document at all, use the certification exam generator.

Build your first AI-103 practice set

Microsoft has not shipped a practice assessment for AI-103. Upload your Foundry notes and generate exam-style questions in under a minute.

Z tej samej rodziny narzędzi