AI-102 Is Retired: AI-102 vs AI-103, Validity and What to Do Next
AI-102 is retired. Microsoft retired the AI-102 exam, which earned the Azure AI Engineer Associate certification, on June 30, 2026, at 11:59 PM Central time, and replaced it with AI-103 (Azure AI Apps and Agents Developer Associate). If you already hold AI-102, it stays valid until its expiration date, but you can no longer renew it.
Below: the key facts, what happens to an AI-102 you already hold, what to do if you were studying for it, and a domain-by-domain AI-102 vs AI-103 comparison built from Microsoft’s two study guides.
Correction, September 2026: earlier versions of this page said AI-102 had not been retired. That was wrong. The page now follows Microsoft’s official pages.
AI-102 retirement at a glance
| Exam | AI-102: Designing and Implementing a Microsoft Azure AI Solution |
|---|---|
| Certification | Microsoft Certified: Azure AI Engineer Associate |
| Status | Retired June 30, 2026, at 11:59 PM Central time. The exam, the certification and its renewal assessment are all retired. |
| Replacement | AI-103: Developing AI Apps and Agents on Azure, which earns Microsoft Certified: Azure AI Apps and Agents Developer Associate |
| Can you still book AI-102? | No |
| Is an AI-102 you already earned still valid? | Yes, until the expiration date in your Microsoft Learn profile |
| Can you renew AI-102? | No. Microsoft retired the renewal assessment. |
| AI-103 price | $165 in the US; priced by country (for example £106 in the UK, €126 in Germany, ₹4,865 in India) |
| AI-103 exam time | 120 minutes (AI-102 gave you 100) |
| Passing score | 700 on both exams |
| Free official practice | Microsoft’s AI-103 practice assessment on AI Skills Navigator (sign-in required) |
Checked against Microsoft Learn’s AI-102 and AI-103 study guides, the Azure AI Engineer Associate and Azure AI Apps and Agents Developer Associate certification pages, Microsoft’s exam pricing and its Skills Hub retirement announcement on October 8, 2026.
Is AI-102 still valid?
Yes, if you earned it. Microsoft says retirement does not revoke or invalidate certifications that were earned. Your Azure AI Engineer Associate stays valid, and stays in the Active section of your Microsoft Learn transcript, until it expires.
What you can’t do is extend it. Associate certifications last one year, and the only way to renew one is the free renewal assessment on Microsoft Learn. Microsoft retired the AI-102 renewal assessment along with the exam, so the expiration date on your profile is now final. Normally an expired certification is earned again by passing its exam, and AI-102 can no longer be taken.
To find your date, sign in to Microsoft Learn, open your profile, choose Certifications and find Azure AI Engineer Associate. Until that date, list it on your resume and LinkedIn as usual. For how long other AI certifications last, see do AI certifications expire?
What replaced AI-102?
AI-103: Developing AI Apps and Agents on Azure. It earns a new credential, Microsoft Certified: Azure AI Apps and Agents Developer Associate, rather than a renamed Azure AI Engineer Associate. Microsoft announced the beta on April 16, 2026, listed the exam as generally available in its June 2026 credentials roundup, and names AI-103 as the exam to take instead of AI-102.
AI-103 costs $165 in the US, runs 120 minutes and needs 700 to pass. It’s offered in English, Chinese (Simplified and Traditional), French, German, Italian, Japanese, Korean, Portuguese (Brazil) and Spanish. The full price breakdown is in our AI-103 exam cost guide.
Seen “AI-103 vs AI-300”? They aren’t two versions of the same thing. AI-300 replaced DP-100 (Azure Data Scientist Associate) and covers running machine learning and generative AI in production. AI-103 is for building AI apps and agents. See DP-100 is retired and the AI-300 study guide.
AI-102 vs AI-103 side by side
| AI-102 (retired) | AI-103 (current) | |
|---|---|---|
| Certification | Azure AI Engineer Associate | Azure AI Apps and Agents Developer Associate |
| Status | Retired June 30, 2026 | Active and generally available |
| Who it’s for | Azure AI engineers building vision, language, knowledge mining and generative AI solutions | Azure AI engineers building agents and AI solutions on Microsoft Foundry |
| Coding languages | Python or C# | Python |
| Exam time | 100 minutes | 120 minutes |
| Price | Can no longer be booked | $165 in the US, priced by country |
| Passing score | 700 on a scale of 1 to 1,000 | |
| Exam languages | The same 10, including English, Japanese, Spanish and Portuguese (Brazil) | |
| Skills outline dated | December 23, 2025 (final version) | April 16, 2026 |
| Renewal | Not possible; renewal assessment retired | Free online renewal assessment on Microsoft Learn each year |
AI-102 vs AI-103 skills, domain by domain
This is how Microsoft’s final AI-102 outline (skills measured as of December 23, 2025) maps onto the AI-103 outline (skills measured as of April 16, 2026). The weights are Microsoft’s published ranges. “No longer named” means a skill isn’t in the AI-103 skills list; it can still come up as background.
| AI-102 domain | AI-103 domain | What changed |
|---|---|---|
| Plan and manage an Azure AI solution (20–25%) | Plan and manage an Azure AI solution (25–30%) | Bigger and built around Foundry: choosing large, small and multimodal models, picking memory and tool services for agents, managing quotas and cost, monitoring drift and grounding quality, keyless credentials and private networking, and governing what agents are allowed to do. |
| Implement generative AI solutions (15–20%) and Implement an agentic solution (5–10%) | Implement generative AI and agentic solutions (30–35%) | Merged into the largest domain: RAG, multistep workflows, evaluating for fabrications and safety, agents with retrieval, function calling and memory, multi-agent orchestration with approval controls, and tracing. AI-102’s prompt flow and fine-tuning bullets are not in the AI-103 outline. |
| Implement computer vision solutions (10–15%) | Implement computer vision solutions (10–15%) | Same weight, new content: generating and editing images and video, multimodal image understanding, captions and alt text, Content Understanding for visual extraction, and defending against prompt injection hidden in images. Custom vision model training and Video Indexer are no longer named. |
| Implement natural language processing solutions (15–20%) | Implement text analysis solutions (10–15%) | Smaller. Entities, summaries and structured JSON output through prompting, translation with Azure Translator or language model flows, and speech as a way to talk to agents. Custom language understanding models and question answering knowledge bases are no longer named. |
| Implement knowledge mining and information extraction solutions (15–20%) | Implement information extraction solutions (10–15%) | Smaller and aimed at grounding: ingesting and indexing documents, images, audio and video; semantic, hybrid and vector search; OCR in RAG ingestion; Content Understanding analyzers. Document Intelligence custom models and Knowledge Store projections are no longer named. |
The pattern: AI-102 asked you to pick and configure individual Azure AI services. AI-103 asks you to build generative apps and agents in Microsoft Foundry and use those services as their tools. Generative AI and agents went from 20–30% of AI-102 to 30–35% of AI-103, and the planning domain, which now covers agent security and governance, grew to 25–30%.
I was studying for AI-102. What now?
Switch to AI-103. Your prep isn’t wasted: Azure Vision, Language, Speech, AI Search and document extraction all come back in AI-103’s vision, text analysis and information extraction domains. Put the extra time where the weight moved.
- Read the AI-103 outline first. Our AI-103 study guide walks through every domain and skill.
- Learn Foundry and agents. Generative AI and agentic solutions are 30–35% of the exam, the largest single domain.
- Move to Python if you used C#. Microsoft’s AI-103 audience profile asks for experience developing apps in Python; C# is no longer mentioned.
- Check the date on your material. Anything written for AI-102 misses the Foundry and agent content, and “AI-103” question sets that are really relabeled AI-102 questions will mislead you. Here’s why AI-103 dumps are a bad bet.
- Test yourself early. Take Microsoft’s free practice assessment on AI Skills Navigator and our free AI-103 practice questions, then follow our AI-103 study plan.
I hold AI-102. Should I take AI-103?
If your job calls for a current Microsoft AI credential, plan to pass AI-103 before your AI-102 expires. AI-103 is also the stepping stone to Microsoft’s new expert credential: Multi-Agent AI Solutions Expert (exam AI-500) requires the AI-103 certification. As an AI-102 holder you already know much of the vision, language and search material, so most of your study time goes to Foundry, agents and Python.
Not sure it pays off? Is AI-103 worth it? weighs it up, and Microsoft AI certifications in 2026 shows where it sits in the new lineup. Coming from fundamentals instead? AI-901 vs AI-103 covers that decision.
Why did Microsoft retire AI-102?
AI-102 went in a 2026 refresh of Microsoft’s cloud and AI certifications. The same announcement retired AI-900 (replaced by AI-901) and DP-100 (replaced by AI-300), plus AZ-204 (replaced by AI-200) and AZ-500 (replaced by SC-500). Microsoft says it is evolving its certifications as fast as roles are changing, and for Azure AI engineers that means generative apps and agents built in Microsoft Foundry. Its renewal FAQ lists the general reasons it retires certifications: relevance, product changes, and changes to how a certification is earned.
Job postings that still ask for AI-102
Hiring posts lag certification changes, so you’ll keep seeing AI-102 in listings for a while. If you hold it, list it as usual with its dates. If you don’t, you can’t take it anymore, so say so in your application: “Azure AI Apps and Agents Developer Associate (AI-103, the successor to AI-102).”
Next step: see where you stand on the exam that replaced AI-102.
Try the free AI-103 practice questions →
Sources: Microsoft Learn’s AI-102 and AI-103 study guides, the Azure AI Engineer Associate and Azure AI Apps and Agents Developer Associate certification pages, Microsoft’s credential expiration policy and renewal FAQ, Microsoft’s exam pricing, and Microsoft’s Skills Hub announcements of the 2026 retirements and new certifications. AI-102, AI-103, Azure and Microsoft Foundry are trademarks of Microsoft Corporation; we reference them descriptively.
Get AI Cert Watch: an email when AI certification exams change
We check the official pages of 80+ AI exams every day. When a price, version, format or retirement date changes, you’ll get a short email. At most one a week, plus the HOW TO // AI newsletter on Thursdays.
Unsubscribe anytime. Privacy policy.
Related guides
- AI-103 Study Guide & Cheat Sheet (Azure AI Apps & Agents Developer)
- Free AI-103 Practice Questions (Azure AI Apps & Agents Developer)
- Is the AI-103 Certification Worth It? An Honest Look at Microsoft’s Newest AI Cert
- How to Pass AI-103: A 3-Week Study Plan for Microsoft’s New Azure AI Exam
- AI-103 Dumps: Why You Don’t Need Them
Frequently asked questions
Is AI-102 retired?
Yes. Microsoft retired the AI-102 exam on June 30, 2026, at 11:59 PM Central time, together with the Azure AI Engineer Associate certification and its renewal assessment. You can no longer book or renew it.
What replaced AI-102?
AI-103: Developing AI Apps and Agents on Azure, which earns Microsoft Certified: Azure AI Apps and Agents Developer Associate. It costs $165 in the US and gives you 120 minutes.
Is AI-102 still valid?
Yes, until its expiration date. Microsoft says retirement does not revoke certifications that were earned, so your Azure AI Engineer Associate stays on your Microsoft Learn transcript until it expires.
Can I renew my AI-102 certification?
No. Microsoft retired the AI-102 renewal assessment with the exam, and that free online assessment is the only way to renew a Microsoft certification. To hold a current Azure AI credential after yours expires, pass AI-103.
Can I still take the AI-102 exam?
No. AI-102 can no longer be booked or retaken. Microsoft points candidates to AI-103 instead.
What is the difference between AI-102 and AI-103?
AI-103 is built around Microsoft Foundry, generative AI apps and agents. Generative AI and agentic solutions are its largest domain at 30–35%, it expects Python rather than Python or C#, and it runs 120 minutes instead of 100.
Is AI-103 harder than AI-102?
Microsoft doesn't publish difficulty. Both are associate exams with a passing score of 700, but AI-103 runs 20 minutes longer and puts its heaviest weight on building agents and generative apps in Foundry.
How much does AI-103 cost?
$165 in the US. Microsoft prices it by country, for example £106 in the UK, €126 in Germany and ₹4,865 in India, before tax. Renewing it each year is free.
