AWS Machine Learning Certification: MLA-C02, and the Exams That Closed
AWS’s ML certification line-up changed twice in 2026. The Specialty is gone, MLA-C01 has left English, and MLA-C02 is in a $75 beta.
The AWS machine learning certification to aim for now is AWS Certified Machine Learning Engineer – Associate, in its new version, MLA-C02. The older Machine Learning – Specialty took its last candidates on 31 March 2026. The current version of the associate exam, MLA-C01, stopped running in English on 28 September 2026. MLA-C02 has been in beta since 29 September at $75, in English only, and AWS’s September announcement gives 14 January 2027 for its general release.
If you are deciding this week, you have three real options: sit the beta, sit MLA-C01 in Japanese, Korean or Simplified Chinese if you work in one of those languages, or wait for the full release. Everything below comes from AWS’s exam pages, its exam guide for MLA-C02 and its Training and Certification blog post of 1 September 2026, all read on 18 October 2026.
What changed in 2026, in date order
- 31 March 2026: last day to sit AWS Certified Machine Learning – Specialty. Anyone who already holds it stays certified for three years after passing.
- 1 September 2026: registration opens for the MLA-C02 beta; the exam guide is published.
- 28 September 2026: last day to sit MLA-C01 in English.
- 29 September 2026: MLA-C02 beta delivery begins.
- 14 January 2027: MLA-C02 general availability in all languages, and MLA-C01 retires everywhere, according to AWS’s blog. The exam page itself still listed GA dates as “TBD” on 18 October, so re-check before you plan around it.
The pattern matters for anyone buying study material. Courses and question banks labelled “MLS-C01” or “MLA-C01” are written for exams you mostly cannot sit any more in English.
MLA-C02: format, fee and who it is for
| MLA-C02 (beta) | MLA-C01 (current) | |
|---|---|---|
| Exam code | ME1-C02 | MLA-C01 |
| Fee | $75 (beta pricing) | $150 |
| Length | 170 minutes | 130 minutes |
| Questions | 85, multiple choice and multiple response | 65, multiple choice and multiple response |
| Languages | English only during beta | Japanese, Korean, Simplified Chinese (English ended 28 Sep 2026) |
| Recommended experience | 1 year with SageMaker AI, Bedrock and other AWS ML services | 1 year with SageMaker and other AWS ML services |
| Validity | 3 years | 3 years |
AWS Certified Machine Learning Engineer – Associate, beta and current versions, from the AWS exam page read 18 October 2026.
AWS says the updated exam reflects a broader ML engineer role, adding generative AI implementation with Bedrock and retrieval-augmented generation, agentic AI workflows, foundation models and large language models, and responsible AI across both traditional ML and GenAI. The domain structure stays the same; the tasks inside the domains changed.
The exam guide for MLA-C02 gives the weights for the general release: Data Preparation for ML and AI 28%, ML Model and Foundation Model Development 24%, Deployment and Orchestration of ML and AI Workflows 24%, and Operating, Monitoring and Securing ML and AI Solutions 24%. At general release it lists 50 scored and 15 unscored questions and a minimum passing score of 720 on a scale of 100 to 1,000. The beta is longer because it carries extra trial items.
How a beta exam works
The $75 is attractive, but read AWS’s beta rules first. On its before-testing policy page AWS says a beta exam is a valid assessment and a pass earns the full certification for three years. Beta exams are discounted, usually open for one to three months, contain more questions than the standard version, and can be taken only once. If you fail the beta you wait for the standard version to become available before you try again. Results take up to five business days.
So the beta suits someone who is ready now. If you would need two attempts, the standard exam, at whatever price AWS sets for it at release, is the realistic plan.
Where it sits among AWS’s AI exams
AWS now has three AI-focused certifications at three levels. They are not a ladder you must climb in order; none of the three requires you to hold another AWS certification first.
| Certification | Level | Fee | Length | Questions | Who AWS aims it at |
|---|---|---|---|---|---|
| AI Practitioner (AIF-C01) | Foundational | $100 | 90 minutes | 65 | People who need AI/ML and generative AI literacy on AWS |
| Machine Learning Engineer – Associate (MLA-C02 beta) | Associate | $75 beta | 170 minutes | 85 | People with a year building ML on SageMaker AI and Bedrock |
| Generative AI Developer – Professional (AIP-C01) | Professional | $300 | 180 minutes | 75 | Developers with 2+ years building production apps and 1 year implementing generative AI |
AWS AI and ML certifications as listed on AWS exam pages, read 18 October 2026.
AI Practitioner is a literacy exam; our AI Practitioner guide covers it. Note that its Italian and German versions were retired after 15 October 2026. Machine Learning Engineer is the hands-on ML exam, now including GenAI. Generative AI Developer – Professional is for people building generative AI applications in production, and AWS expects two years or more of building production applications plus a year of hands-on generative AI work.
If your job is training, tuning and running models, MLA-C02 is the match. If it is wiring foundation models into products, the GenAI Developer exam is closer, though it costs four times the beta fee and assumes more experience.
What happened to the ML Specialty
The Machine Learning – Specialty was AWS’s advanced ML exam at $300 for 180 minutes and 65 questions, aimed at people with two or more years running ML or deep learning workloads on AWS. AWS retired it with a last sitting on 31 March 2026. People who passed it keep it for three years from their own pass date; AWS also said those holders could recertify by retaking it on or before 31 March 2026, which has now passed.
There is no direct replacement at specialty level. The nearest successors are MLA-C02 for ML engineering and AIP-C01 for generative AI development. Search results still rank “AWS machine learning specialty certification” pages written before the retirement, which is the main reason to check the date on anything you read.
Preparation: free first, then paid
AWS’s preparation page lists what costs nothing on AWS Skill Builder: the Official Practice Question Sets (20 questions each, written to mirror the real exam’s style), the two-hour Exam Prep digital courses, and AWS Cloud Quest. The MLA page links a four-step plan that starts with the exam guide and the free question set.
Paying for Skill Builder (AWS’s pricing page showed $29 monthly and $449 annually on 18 October 2026) adds the Official Practice Exam with exam-style scoring and the longer exam-prep courses with labs. The MLA page also points to an AWS Certification Official Pretest for finding gaps before you start. Our Skill Builder breakdown shows exactly what the subscription unlocks.
Outside AWS, practice libraries such as Whizlabs and courses from KnowledgeHut or Simplilearn are options for more volume. Before you pay, check that the course is labelled MLA-C02: a C01 course will miss the generative AI, agentic and Bedrock material AWS added.
Is it worth it?
There is no official pass rate for MLA, and AWS prints no pay figures for it either; the numbers training companies attach to it are theirs, not AWS’s, so they are not here. What it demonstrably gives you is a current, dated proof that you can build and operate ML systems on AWS, valid for three years, and a 50% discount on your next AWS exam, which AWS’s exam pages say you receive once you hold one AWS certification.
It is a sensible step if your job already involves SageMaker AI or Bedrock, or will soon. If you have never deployed a model, start with hands-on work or the AI Practitioner exam instead; the MLA exam assumes a year of practice. The wider AWS certification path places it among the other associate exams, and our page on whether AWS certification is worth it covers the general case.
A six-week plan for the beta
If you decide to sit the beta while it is open, work back from a booked date rather than studying open-ended. Spend the first fortnight on the exam guide, reading each task statement and marking the ones you have never done in a real account. Spend weeks three and four on the gaps, using the free Exam Prep course and hands-on time in SageMaker AI and Bedrock; the additions AWS lists, retrieval-augmented generation and agentic workflows, reward building one small working example each. Use week five for the free Official Practice Question Set and any timed practice you choose to pay for. Keep week six for review. With 85 questions in 170 minutes, you have two minutes a question, so timed practice matters more than on a shorter exam.
Edge cases
You passed MLA-C01 before 28 September. Your certification stays active for its full three years and you do not need to sit MLA-C02. When renewal comes, AWS’s recertification page lists three routes for this certification: pass the latest version of the exam, pass the Generative AI Developer – Professional exam, or extend it by a year through AWS Skill Builder activities, which need a subscription.
You hold the ML Specialty. It runs for three years from when you earned it, and with the exam closed it can no longer be renewed by retaking it. If you want a current AWS ML credential after it lapses, one of the live exams above is the way to get it.
You failed the beta. You cannot retake the beta. Wait for the generally available version.
You prefer Japanese, Korean or Simplified Chinese. MLA-C01 stays open in those languages until MLA-C02’s general release, at $150.
AWS machine learning certification questions
What is the AWS machine learning certification now?
Is the AWS Machine Learning Specialty retired?
How much does the AWS ML certification cost?
Is there an AWS MLOps certification?
Does AWS publish a salary for the ML certification?
This guide sits on one rung. The whole AWS certification path, level by level shows where it falls, what the next exam costs, and which exams AWS has dated for retirement. Exam facts here come from AWS’s own pages on 24 September 2026; check the exam page again before you pay, because versions and prices change.