The short version
Take AI-103 if your work is the AI itself: models, agents, retrieval, vision, speech, and Microsoft Foundry. Take AI-200 if your work is the cloud plumbing around AI apps: containers, vector-enabled databases, messaging, Functions, secrets, and monitoring. Take both if your job spans the two; there is no required order in Microsoft’s pages. Facts below come from Microsoft’s study guides and certification pages, checked October 22, 2026.
Side-by-side
| Item | AI-103 | AI-200 |
|---|---|---|
| Exam name | Developing AI Apps and Agents on Azure | Developing AI Cloud Solutions on Azure |
| Credential | Azure AI Apps and Agents Developer Associate | Azure AI Cloud Developer Associate |
| Level | Intermediate (associate) | Intermediate (associate) |
| Role focus | Azure AI engineer building agents and AI solutions with Microsoft Foundry | Developer responsible for back-end services and components of AI solutions |
| Skills version | Measured as of April 16, 2026 | Measured as of October 7, 2026 |
| Time | 120 minutes | 120 minutes |
| Pass mark | 700 of 1,000 | 700 of 1,000 |
| Delivery | Proctored, Pearson VUE | Proctored, Pearson VUE |
| Languages listed | 10 | 13 |
| Questions | Not published | Not published |
| Expected skills | Python, general and generative AI, Azure services | Azure SDKs, data services, monitoring, messaging, vector databases, Python, containers |
| Validity and renewal | One year, free renewal assessment | One year, free renewal assessment |
What each exam covers
AI-103 domains: plan and manage an Azure AI solution (25-30%), implement generative AI and agentic solutions (30-35%), computer vision (10-15%), text analysis (10-15%), information extraction (10-15%).
AI-200 domains: containerized solutions (20-25%), AI solutions with Azure data management services (30-35% in the skills-at-a-glance list), connecting to and consuming Azure services (20-25%), and securing, monitoring, and troubleshooting (20-25%). Microsoft’s change log says the data management share increased on October 7, 2026. The detailed heading on the page still showed 25-30% when we checked, so confirm the figure in the study guide.
The first exam asks “what should the AI part do, and how should it behave?” The second asks “where does it run, where does its data live, and how do we keep it secure and observable?”
Where they overlap
- Both are associate-level, 120-minute, 700-to-pass, Pearson VUE exams with the same retake and renewal rules.
- Both expect Python.
- Both touch retrieval over vector data. AI-103 treats it as search and grounding configuration (semantic, hybrid, and vector search). AI-200 treats it as database work (Cosmos DB, PostgreSQL with pgvector, Azure Managed Redis).
- Both include security basics, though AI-103 frames them as managed identity and private networking for AI services, while AI-200 adds Key Vault, App Configuration, and tracing with OpenTelemetry.
- Both mention monitoring: model and agent signals in AI-103, logs, metrics, and KQL in AI-200.
Where they differ
- Only AI-103 includes computer vision, speech, text analysis with language models, responsible AI tooling, and agent design.
- Only AI-200 includes Container Registry, App Service, Container Apps, Kubernetes, Service Bus, Event Grid, and Azure Functions.
- AI-103 is closer to “using AI services well”. AI-200 is closer to “classic Azure developer skills with vector databases added”.
How to choose
- You design prompts, agents, and retrieval; someone else runs the infrastructure: AI-103.
- You deploy, secure, and operate the services AI apps rely on: AI-200.
- You do both: start with the one closer to your current project, because hands-on knowledge shortens study time. The second exam then fills the other half.
- You are new to Azure: neither is a first exam. Microsoft expects practical experience for both; see each prerequisites page.
- You studied the older Azure Developer or Azure AI Engineer certifications: Microsoft’s pages mark both as retired. We found no statement on Microsoft’s pages that either new exam is a direct replacement, so compare the skills lists yourself.
Should you take both?
If you build full AI products on Azure, the pair covers a broad slice of the stack. Study time overlaps little, since the skills lists differ, so plan two separate efforts. Each credential needs its own yearly renewal assessment, which is free.
More guides for AI-103
- AI-103 Exam Format, Scoring, and Retakes
- AI-103 Study Plan: 4-Week and 8-Week Schedules
- AI-103 Sample Questions With Answers
- Is the AI-103 Exam Hard? An Honest Look
- AI-103 Prerequisites and Skills Self-Check
- AI-103 Renewal and Retakes Explained
- AI-103 Cheat Sheet: Key Facts to Review
- AB-620 vs AI-103: Which Exam First?
Next step
Read the full overviews at AI-103 and AI-200, then practice with the AI-103 practice tests or the AI-200 practice tests. For deeper reading on the first exam, see domains explained and Is AI-103 worth it?.
Frequently asked questions
Which one is harder?
It depends on your background. Developers strong in cloud services but new to generative AI find AI-103 harder; AI specialists with little infrastructure experience find AI-200 harder.
Do I need AI-200 before AI-103, or the other way around?
Microsoft’s pages we checked list no prerequisite certification for either.
Are the pass marks different?
No. Both use 700 on a scale of 1 to 1,000.