AWS Machine Learning Specialty Certification Guide

The AWS Machine Learning Specialty (MLS-C01) retired in March 2026. What the exam covered, what happens to existing holders, and which AWS ML certification to take instead.

Daniel Carter17 August 2026 · 8 min read

Ask AIChatGPT Claude Gemini Copilot Perplexity Grok

Signpost showing the retired AWS ML Specialty path and the three current AWS AI certification paths

If you are planning to sit the AWS Certified Machine Learning – Specialty (MLS-C01), stop before you build a study plan: the exam is retired. The last day to take MLS-C01 was 31 March 2026, and as of August 2026 it can no longer be booked or renewed. Machine learning validation on AWS now runs through three newer certifications — AWS Certified AI Practitioner (AIF-C01), AWS Certified Machine Learning Engineer – Associate (MLA-C01, with an MLA-C02 update opening for registration on 1 September 2026), and the Professional-level AWS Certified Generative AI Developer (AIP-C01).

This guide covers all of it in one place: what the Machine Learning Specialty was, exactly what its retirement means if you hold it, and how to choose your replacement exam. For where an ML certification slots into a broader multi-cert plan, the AWS Certification Roadmap for 2026 handles sequencing across the whole portfolio.

What was the AWS Certified Machine Learning – Specialty?

The Machine Learning – Specialty was AWS's single, monolithic machine learning credential for years: a Specialty-tier exam validating that a practitioner could design, build, train, tune and deploy ML models on AWS, with Amazon SageMaker at the centre of the tested workflow alongside the data engineering and exploratory analysis work that surrounds real ML projects.

In its final form, MLS-C01 ran to 65 questions in 180 minutes, cost $300 USD, and required a scaled passing score of 750 — the standard Specialty threshold. Like every AWS exam it had no formal prerequisites, carried a three-year validity period, and was delivered at Pearson VUE test centres or via online proctoring.

Two things made it distinctive while it lived. It was one of the longest AWS exams by seat time, and it was unusually broad: a single paper stretching from data preparation through modelling theory to production deployment. That breadth is precisely what AWS unbundled when it restructured its AI portfolio.

Why did AWS retire MLS-C01?

AWS retired the Machine Learning Specialty as part of a deliberate expansion of its AI and machine learning certification portfolio, announced on the AWS Training and Certification blog. Rather than one Specialty exam covering everything from fundamentals to deployment, the portfolio now splits ML validation by role and depth:

Foundational understanding of AI/ML concepts and services → AI Practitioner (AIF-C01), launched October 2024.
Hands-on ML engineering — building, operationalising and maintaining ML workloads → Machine Learning Engineer – Associate (MLA-C01), launched October 2024.
Production generative AI development — integrating foundation models into applications → Generative AI Developer – Professional (AIP-C01), beta from November 2025, standard registration open as of March 2026.

Seen against that line-up, the old Specialty had become redundant: its audience was carved up among three better-targeted exams. The retirement follows the same pruning pattern as the AWS Database Specialty, Data Analytics Specialty and SAP on AWS Specialty, all retired in April 2024.

The dates that matter

Date	Event
October 2024	AIF-C01 and MLA-C01 launch, introducing new ordering, matching and case-study question types
November 2025	AIP-C01 (Generative AI Developer – Professional) opens beta registration at $150
March 2026	AIP-C01 standard registration open
31 March 2026	Last day to take MLS-C01
1 September 2026	MLA-C02 registration opens
28 September 2026	Last day to take MLA-C01 in English (Korean, Japanese and Simplified Chinese versions continue until MLA-C02 reaches those languages)

I already hold the ML Specialty — what happens to my certification?

Nothing dramatic, and nothing immediately. Your certification remains active for the full three years from the date you earned it. AWS retirements never revoke existing credentials, so an MLS earned in, say, late 2025 stays valid into 2028 and can appear on your CV as an active AWS certification throughout.

What changes is renewal. Specialty certifications renew only by retaking the Specialty exam, and there is no longer an exam to retake — so when your three years expire, the certification lapses. Practical implications:

Plan a successor exam before your expiry date, not after. The natural continuation for most MLS holders is the Machine Learning Engineer – Associate, or the Generative AI Developer – Professional for those working with foundation models in production.
Use your discount. While your MLS is still active, you qualify for AWS's 50% discount voucher on your next exam — a concrete reason to book the successor before, rather than after, the old credential expires.
Keep the credential listed while valid. A retired exam's certification is not a devalued one; it still attests to what you passed, until its printed expiry.

Which AWS certification should you take instead of MLS-C01?

There is no universal answer — the right exam depends on what you actually do with machine learning. Use the profiles below as a decision framework.

If you build and ship ML systems: Machine Learning Engineer – Associate

The closest working replacement for the Specialty's core audience. MLA validates hands-on ML engineering on AWS — AWS suggests at least a year of experience with Amazon SageMaker and related services. Format: 65 questions, 130 minutes, $150 USD, passing score 720 scaled, three-year validity. Along with AI Practitioner it was the first AWS exam to include the new ordering, matching and case-study question types, so expect more varied question mechanics than the old all-multiple-choice Specialty.

One timing decision is unavoidable right now: MLA is mid-update. Registration for MLA-C02 opens on 1 September 2026, and 28 September 2026 is the last day to sit MLA-C01 in English. If you are nearly ready, booking MLA-C01 before that cut-off lets you use materials written for the current version; if you are months away, plan for MLA-C02 and rely on AWS's official exam guide for it once published — the content changes have not been detailed publicly, so treat any third-party claims about MLA-C02's content with suspicion.

If you work around ML rather than in it: AI Practitioner

AIF-C01 is a Foundational exam — 65 questions, 90 minutes, $100 USD, passing score 700 — aimed at people who are familiar with AI/ML solutions on AWS without necessarily building them: product managers, analysts, sales engineers, project leads. It is offered in 12 languages. A former MLS candidate will find it far shallower than what they were preparing for; choose it only if your goal was vocabulary and fluency rather than engineering proof.

If your work is generative AI in production: Generative AI Developer – Professional

AIP-C01 is the new top of the AI track: a Professional-level certification validating that you can integrate foundation models into production applications, including work with services such as Amazon Bedrock. Standard price is $300 USD with the usual three-year validity, and AWS recommends two or more years of cloud experience plus at least a year of hands-on generative AI work. For senior MLS holders whose day job has shifted toward large-model applications, this is the more forward-looking replacement.

If your bottleneck is data, not models: Data Engineer – Associate

Plenty of "ML" job descriptions are mostly pipeline work. If your gaps are ingestion, transformation and data infrastructure rather than model training, the AWS Data Engineer Associate Certification Guide covers DEA-C01, the Associate exam that effectively absorbed AWS's data-analytics territory.

Successor exams compared

Factor	AI Practitioner (AIF-C01)	ML Engineer – Associate (MLA-C01/C02)	Generative AI Developer – Professional (AIP-C01)
Level	Foundational	Associate	Professional
Format	65 questions, 90 min	65 questions, 130 min	See official exam guide
Cost (USD, as of 2026)	$100	$150	$300
Passing score (scaled)	700	720	Confirm in official exam guide
Prerequisites	None	None	None
Recommended background	Familiarity with AI/ML on AWS	1+ year with SageMaker and related services	2+ years cloud, 1+ year hands-on generative AI
Best for	Non-builders needing AI fluency	ML engineers shipping models	Developers productionising foundation models
Renewal	3 years	3 years	3 years

All three, like every AWS exam, have no formal prerequisites and are bookable at Pearson VUE centres or online.

How to prepare for whichever successor you choose

The study mechanics that would have served you for MLS-C01 transfer almost intact:

Anchor on the official exam guide for your chosen exam and turn its task statements into a personal gap list.
Prioritise hands-on work in the services the guide names — reading about a training pipeline and having debugged one are different exam skills, and the scenario-style questions test the second.
Rehearse the new question types if you sit MLA or AIF: ordering, matching and case-study items reward genuinely understanding a workflow's sequence, which answer-memorisation cannot fake.
Benchmark with timed practice and read the results diagnostically — a domain where you miss repeatedly is a study target, not a re-attempt target. ExamPractice's amazon exams hub collects AWS practice materials, with free sample questions per exam and a timed simulation mode for subscribers, which is a sensible final checkpoint before you spend the exam fee.

If you have old MLS-C01 study materials, mine them selectively: the modelling fundamentals and SageMaker workflow content still teaches real skills, but the exam blueprint, domain weightings and question style they target no longer exist.

Frequently asked questions

Can I still book the AWS Machine Learning Specialty anywhere?

No. The final sitting date was 31 March 2026, and the retirement is global. Any site claiming to sell MLS-C01 bookings after that date is out of date.

Does the retirement make my existing MLS certification worthless?

No. It stays active until its normal three-year expiry, and it certifies the same competence it always did. It simply cannot be renewed once expired.

Is the ML Engineer Associate easier than the ML Specialty was?

They sit at different tiers — Associate (720 passing score, 130 minutes) versus Specialty (750, 180 minutes) — and MLA has a narrower, more engineering-focused scope than the sprawling MLS blueprint. AWS does not publish pass rates for either, so treat any claimed difficulty statistics as unverifiable; the honest comparison is scope and level, not numbers.

Should I wait for MLA-C02 or take MLA-C01 now?

If you can be ready before 28 September 2026 (the last English MLA-C01 date), sitting the current version means studying against a stable, well-documented blueprint. If not, wait for MLA-C02 — registration opens 1 September 2026 — and base your preparation on its official exam guide rather than speculation about what changed.

Your next move after the Specialty

The Machine Learning Specialty had a good run as AWS's single badge for ML competence, but its retirement is best read as a signal rather than a loss: AWS now expects ML validation to match your role. Builders take the Machine Learning Engineer – Associate, foundation-model developers aim at the Generative AI Developer – Professional, and non-engineers get honest value from AI Practitioner. Pick the lane that matches your actual work, check the current details on the official AWS certification pages before booking — this corner of the portfolio has changed every year since 2024 — and put your study hours into the exam that will still exist when you are ready to sit it.

Exam facts in this guide were checked against official certification-provider pages on 30 August 2026. Fees, exam codes and policies change — confirm on the provider’s own site before you book.