Career upgrade: Learn practical AI skills for better jobs and higher pay.
Level up
All Practice Exams

100+ Free AI-103 Azure AI Apps and Agents Developer Practice Questions

Pass your Microsoft Certified: Azure AI Apps and Agents Developer Associate (Exam AI-103) exam on the first try — instant access, no signup required.

✓ No registration✓ No credit card✓ No hidden fees✓ Start practicing immediately
100+ Questions
100% Free
1 / 100
Question 1
Score: 0/0

Which Foundry capability connects an agent to external content such as search indexes and knowledge stores so it can ground responses?

A
B
C
D
to track
2026 Statistics

Key Facts: AI-103 Azure AI Apps and Agents Developer Exam

$165

Exam Fee (USD)

Microsoft / Pearson VUE

~100 min

Exam Duration

Microsoft

700/1000

Passing Score

Microsoft Learn

40-60

Question Count

Microsoft

30-35%

Generative AI & Agents Weight

AI-103 study guide

1 year

Credential Validity (free renewal)

Microsoft Learn

As of the April 16, 2026 study guide, Microsoft Exam AI-103 (Developing AI Apps and Agents on Azure) costs $165 USD, runs about 100 minutes with roughly 40-60 questions, and requires a score of 700 out of 1000 to pass. It is delivered through Pearson VUE and renews free online once per year. The largest skill area is Implement generative AI and agentic solutions at 30-35%, followed by Plan and manage an Azure AI solution at 25-30%, with computer vision, text analysis, and information extraction each at 10-15%.

Sample AI-103 Azure AI Apps and Agents Developer Practice Questions

Try these sample questions to test your AI-103 Azure AI Apps and Agents Developer exam readiness. Each question includes a detailed explanation. Start the interactive quiz above for the full 100+ question experience with AI tutoring.

1Which Azure platform is the primary environment for building, deploying, and managing generative AI apps and agents on the AI-103 exam?
A.Azure DevOps
B.Microsoft (Azure AI) Foundry
C.Azure Synapse Analytics
D.Azure Machine Learning Designer
Explanation: Microsoft Foundry (Azure AI Foundry) is the unified platform candidates use to build, manage, and deploy agents and generative AI solutions. The AI-103 audience profile defines the role as an Azure AI engineer who takes advantage of Foundry. The other services exist but are not the central authoring environment for this exam.
2An organization needs a generative model that can reason over both text and images in a single request. Which model type should you select in the Foundry model catalog?
A.A small language model optimized for edge devices
B.A multimodal model
C.A text-only embedding model
D.A speech-to-text model
Explanation: Multimodal models accept and reason over more than one input type, such as text combined with images, in a single prompt. The AI-103 study guide calls out choosing multimodal models for multimodal processing tasks. Embedding and speech models serve narrower, single-purpose roles.
3When is choosing a small language model (SLM) over a large language model the most appropriate decision?
A.When the workload demands the broadest possible general-knowledge reasoning
B.When you need lower cost and latency for a focused, well-scoped task
C.When you must generate high-resolution images from prompts
D.When you require speech translation across many languages
Explanation: Small language models trade some breadth of reasoning for reduced cost, lower latency, and smaller footprint, which suits narrow, well-defined tasks. The AI-103 study guide lists choosing among LLMs, small language models, and multimodal models per task. Broad reasoning favors a larger model, and image or speech tasks need specialized models.
4Which retrieval approach combines keyword (lexical) matching with vector similarity to improve grounding relevance?
A.Pure full-text search
B.Hybrid search
C.Exact-match filtering only
D.Round-robin scoring
Explanation: Hybrid search runs both keyword and vector queries and fuses the results, capturing exact-term matches and semantic similarity together. The AI-103 study guide lists configuring semantic, hybrid, and vector search for grounding. Pure full-text or exact-match alone miss semantic matches.
5An agent must recall context across multiple turns of a long conversation. Which integration capability addresses this requirement?
A.Conversation memory
B.Static prompt templating only
C.Image inpainting
D.Optical character recognition
Explanation: Conversation memory lets an agent retain and reuse prior turns so it can maintain context across a session. The AI-103 study guide lists choosing memory, tool, and knowledge integration services for agents and building agents that integrate conversation memory. The other options serve image or document tasks.
6Which credential approach best satisfies a security requirement to avoid storing API keys in application configuration when accessing Azure AI services?
A.Embed the key in environment variables
B.Use managed identity with keyless authentication
C.Hard-code the key in source control
D.Share a single key across all environments
Explanation: Managed identity provides an Azure AD identity to a resource so it can obtain tokens without any stored secret, which is the keyless credential pattern. The AI-103 study guide explicitly lists managed identity and keyless credentials under security configuration. Storing or hard-coding keys violates that guidance.
7To restrict an Azure AI Foundry resource so it is reachable only from within a virtual network and not the public internet, which feature should you configure?
A.A public endpoint with an allow-all rule
B.Private networking with private endpoints
C.A shared access signature on blob storage
D.Cross-origin resource sharing
Explanation: Private endpoints place the service on a private IP inside the virtual network and disable public access, satisfying network isolation requirements. The AI-103 study guide lists private networking as a security configuration. The other options either keep the resource public or address unrelated concerns.
8Which Azure RBAC concept lets you grant a service principal only the specific permissions it needs on a Foundry project?
A.Assigning the Owner role for convenience
B.Applying a least-privilege role assignment
C.Disabling Azure AD authentication
D.Granting access through a shared key only
Explanation: Least-privilege role assignments grant only the permissions required, reducing risk if the identity is compromised. The AI-103 study guide lists role policies under security configuration. Owner is overly broad, disabling Azure AD weakens security, and key-only access bypasses role-based control.
9A team wants to cap monthly spend and prevent a single model deployment from consuming all available capacity. Which controls should they configure?
A.Quotas and rate limits
B.Content safety filters
C.Vector index refresh schedules
D.Alt-text generation rules
Explanation: Quotas cap allocated capacity and rate limits throttle request volume, together managing cost footprint and preventing capacity exhaustion. The AI-103 study guide lists managing quotas, scaling, rate limits, and cost footprints for model and agent workloads. The other options address safety, search, or accessibility, not capacity governance.
10Which signal should you monitor to detect when a deployed model's input data distribution has shifted away from what it was validated against?
A.Token pricing tier
B.Model drift
C.Region availability
D.Subscription quota name
Explanation: Model drift describes a change in input or output distribution over time that can degrade quality, so it must be monitored. The AI-103 study guide lists monitoring model performance, drift, safety events, and grounding quality. The other options are billing or configuration attributes, not quality signals.

About the AI-103 Azure AI Apps and Agents Developer Exam

Exam AI-103 validates an Azure AI engineer's ability to build, manage, and deploy generative AI apps and agents using Microsoft (Azure AI) Foundry. The blueprint spans planning and managing Azure AI solutions, implementing generative AI and agentic solutions with RAG and tools, computer vision, text analysis, and information extraction. AI-103 is the successor to AI-102 and entered beta in April 2026.

Questions

50 scored questions

Time Limit

100 minutes

Passing Score

700/1000

Exam Fee

$165 (Microsoft)

AI-103 Azure AI Apps and Agents Developer Exam Content Outline

25-30%

Plan and manage an Azure AI solution

Choose appropriate Foundry services and models for generative AI and agents, set up and deploy AI solutions, integrate CI/CD, manage quotas, scaling, rate limits and cost, secure with managed identity, private networking, keyless credentials and role policies, and implement responsible AI guardrails, evaluators, auditing, and oversight controls.

30-35%

Implement generative AI and agentic solutions

Deploy and consume LLMs, small, code, and multimodal models, implement RAG, design tool-augmented and multistep reasoning workflows, build agents with roles, tool schemas, function calling and memory, orchestrate multi-agent and approval-gated flows, and optimize with prompt engineering, reflection, tracing, and token analytics.

10-15%

Implement computer vision solutions

Generate and edit images and video from prompts and reference media, build multimodal understanding with Content Understanding including captioning, visual Q&A, alt-text, and object detection, and apply responsible AI such as visual content filters, indirect prompt-injection mitigation, and visual policy enforcement.

10-15%

Implement text analysis solutions

Use language models and Foundry Tools to extract entities, topics, summaries, and structured JSON, detect sentiment, tone, safety issues and sensitive content, translate text with Azure Translator, and build speech-to-text, text-to-speech, custom speech, and speech-translation flows for agents.

10-15%

Implement information extraction solutions

Build retrieval and grounding pipelines that ingest and index documents, images, audio, and video, configure semantic, hybrid, and vector search, apply OCR and enrichment skills, connect retrieval to agent tools, and extract document content into clean structured or markdown outputs with Content Understanding.

How to Pass the AI-103 Azure AI Apps and Agents Developer Exam

What You Need to Know

  • Passing score: 700/1000
  • Exam length: 50 questions
  • Time limit: 100 minutes
  • Exam fee: $165

Keys to Passing

  • Complete 500+ practice questions
  • Score 80%+ consistently before scheduling
  • Focus on highest-weighted sections
  • Use our AI tutor for tough concepts

AI-103 Azure AI Apps and Agents Developer Study Tips from Top Performers

1Spend the largest share of prep time on generative AI and agentic solutions, since it is the heaviest domain at 30-35%.
2Know Microsoft (Azure AI) Foundry end to end: model catalog, deployments, agent definitions, tool schemas, function calling, and the Foundry SDK project connection.
3Master RAG concepts and Azure AI Search retrieval modes; be able to choose between keyword, vector, semantic, and hybrid search for a given scenario.
4Practice responsible AI topics across every domain: content safety filters, evaluators, groundedness and fabrication detection, auditing, approval flows, and indirect prompt-injection mitigation.
5Review security patterns precisely: managed identity, keyless credentials, private networking, and least-privilege role policies appear throughout the plan-and-manage domain.
6Do timed mixed-question sets so the roughly 100-minute, 40-60 question pace and scenario-based items feel routine before exam day.

Frequently Asked Questions

What are the current official exam facts for AI-103?

Microsoft lists Exam AI-103 at $165 USD with about 40-60 questions in roughly 100 minutes, and a passing score of 700 out of 1000. The exam is delivered through Pearson VUE and is the successor to AI-102, having entered beta in April 2026.

What skills does AI-103 measure and how are they weighted?

Plan and manage an Azure AI solution is 25-30%, Implement generative AI and agentic solutions is 30-35%, and computer vision, text analysis, and information extraction are each 10-15%. The generative AI and agentic domain is the largest single area.

Does AI-103 require prior experience?

There are no formal prerequisites, but Microsoft recommends experience developing apps with Python and familiarity with general AI, generative AI, and Azure services. The target role is an Azure AI engineer building agents and AI solutions with Foundry.

How is AI-103 different from AI-102?

AI-103 is the successor to AI-102 and centers on building AI apps and agents in Microsoft (Azure AI) Foundry, including agentic workflows, multi-agent orchestration, RAG, and responsible AI. It went into beta in April 2026.

How long is the certification valid after you pass?

Microsoft associate certifications expire one year after you earn them. You renew for free by passing a short online assessment on Microsoft Learn during the renewal window.

What is the best way to prepare for AI-103?

Start with the official Microsoft Learn study guide and learning paths, then drill mixed practice questions across all five skill areas. Because generative AI and agentic solutions is the largest domain, spend the most time on Foundry agents, RAG, tools, and responsible AI.