AIF-C01 vs AI-901 is one of the most common questions from people starting in AI. Both are entry level AI certifications. AIF-C01 is the AWS Certified AI Practitioner. AI-901 is Microsoft Azure AI Fundamentals, which replaced AI-900 in 2026. They cover similar ideas, but they test them in different ways and lead to different next steps. This comparison helps you choose the one that fits your goals.
Quick answers
- Which is easier? Both are entry level. AIF-C01 needs no coding. AI-901 expects you to read short Python snippets that use Microsoft Foundry.
- Which has more weight on generative AI? Both lean heavily on generative AI, foundation models and responsible AI.
- Which should I take first? The one for the cloud your current or target employer uses. If you have no preference, see the decision guide below.
- Is AI-900 still available? No. AI-900 retired on June 30, 2026, and AI-901 replaced it.
- Do I need both? Rarely. One entry level AI certification is enough. Move to an associate level exam next.
AIF-C01 vs AI-901 at a glance
| AIF-C01 (AWS) | AI-901 (Microsoft) | |
|---|---|---|
| Full name | AWS Certified AI Practitioner | Microsoft Azure AI Fundamentals |
| Level | Foundational | Fundamentals |
| Exam length | 65 questions (50 scored), 90 minutes | See the AI-901 exam page on Microsoft Learn |
| Passing score | 700 / 1000 | 700 / 1000 |
| Skills outline date | October 29, 2026 | April 15, 2026 |
| Main platform | Amazon Bedrock and SageMaker AI | Microsoft Foundry |
| Coding | None required | Read basic Python |
| Validity | Renew to stay active (renewal adds three years) | Fundamentals certifications do not expire |
AIF-C01 domains and weights
The AWS exam guide (as of October 29, 2026) uses five domains:
| Domain | Weight |
|---|---|
| Fundamentals of AI and ML | 20% |
| Fundamentals of GenAI | 24% |
| Applications of Foundation Models | 28% |
| Guidelines for Responsible AI | 14% |
| Security, Compliance, and Governance for AI Solutions | 14% |
AIF-C01 asks you to choose the right approach and the right AWS service. Expect questions on when to use RAG, fine-tuning or continued pre-training, which managed AI service fits a use case (Textract, Comprehend, Rekognition, Transcribe and others), and how Bedrock Guardrails, SageMaker Clarify and AWS security services support responsible AI.
AI-901 domains and weights
The AI-901 skills outline (as of April 15, 2026) has just two domains:
| Domain | Weight |
|---|---|
| Identify AI concepts and capabilities | 40 to 45% |
| Implement AI solutions by using Microsoft Foundry | 55 to 60% |
More than half of AI-901 is about doing things in Microsoft Foundry: deploying a chat model, testing prompts and temperature, building an agent with file search, and using text, speech, vision and Content Understanding features. The concepts domain covers the six Microsoft responsible AI principles, tokens, prompts and matching workloads to the right capability.
The real difference: choosing vs building
Both exams test generative AI concepts. The difference is the angle.
- AIF-C01 is about choosing. You read a business scenario and pick the approach and service. It suits product managers, analysts, sales engineers, consultants and anyone who works with AI projects without writing the code.
- AI-901 is about building, lightly. You need to know what happens when you deploy a model, call it from Python and add tools to an agent. It suits developers and technical beginners who want to build on Azure.
Which should you take first? A simple decision guide
- Your employer or target employer uses AWS: take AIF-C01.
- Your employer uses Microsoft 365 and Azure: take AI-901.
- You do not write code and do not plan to: AIF-C01 is the more comfortable fit.
- You want to build AI apps soon: AI-901, then move to AI-103.
- Still unsure: look at ten job posts for the role you want and count which cloud appears more. Our guide on which AI certification to get first walks through this in more detail.
How to prepare for each
Preparing for AIF-C01
Hands on time in the Amazon Bedrock console matters more than reading. Enable model access, try a few models in the playground, change temperature and max tokens, build a Bedrock Knowledge Base over some PDFs, and create a Guardrail with a denied topic. Learn the classic metrics too: accuracy, precision, recall and F1 for classification, ROUGE for summaries and BLEU for translation.
Preparing for AI-901
Try every feature yourself in the Microsoft Foundry portal, then repeat it in a few lines of Python. Deploy a chat model, write a small client, build an agent with file search, run sentiment and entity detection on sample text, and create a Content Understanding analyzer. Finish with the free official practice assessment on Microsoft Learn.
Common mistakes with either exam
- Using outdated material. Both exams changed recently. AI-901 replaced AI-900 in 2026 and is built around Microsoft Foundry. Check the outline date on any course or book before you buy it.
- Skipping responsible AI. It is 14% of AIF-C01 and part of the concepts domain in AI-901. Learn the principles with a real example for each, not only the names.
- Confusing RAG with fine-tuning. RAG brings in current documents at question time. Fine-tuning changes how the model behaves using labeled examples. Many scenario questions turn on this difference.
- Reading without trying. Ten minutes in the Bedrock or Foundry playground teaches temperature, max tokens and system prompts better than an hour of reading.
Try one original question from each
These are original practice examples written for this article, not real exam questions.
A company wants a chatbot that answers employee questions from internal policy documents and cites the source. The documents change every week. Which approach fits best on AWS?
Correct answer: B
RAG retrieves the latest documents at question time and can return citations, with no retraining when documents change.
You want a model in Microsoft Foundry to give consistent, factual answers with less variation between runs. Which setting should you change?
Correct answer: A
A lower temperature makes the output more deterministic. Max tokens only limits answer length.
Frequently asked questions
Is AIF-C01 harder than AI-901?
Neither is hard if you prepare, but they feel different. AIF-C01 has longer business scenarios. AI-901 has more hands on detail about Foundry. People who never code usually find AIF-C01 more comfortable.
What comes after AIF-C01?
Developers often move to the AWS Generative AI Developer Professional (AIP-C01). If you need broader AWS skills first, the Solutions Architect Associate is a strong next step.
What comes after AI-901?
The natural next step is AI-103, which replaced AI-102. Read what changed from AI-102 to AI-103.
Can I still use AI-900 study material for AI-901?
Only partly. AI-901 is built around Microsoft Foundry, generative AI and agents, so older AI-900 material misses a large part of the exam.
Practice before you choose
Each exam page has free practice questions with full explanations. Try a few on the AIF-C01 page and the AI-901 page and see which style suits you. The complete guide for either exam is $19 and includes a full question bank, every option explained, a four week study plan and a cheat sheet. Planning to take both, or more? The All-Access Pass includes every guide.