MLA-C01 and MLA-C02 are two versions of the same AWS Certified Machine Learning Engineer Associate exam. The last day to take MLA-C01 in English is September 28, 2026, and MLA-C02 beta delivery starts September 29, 2026. MLA-C02 keeps the four-domain, 65-question format and adds generative AI, agentic AI, and foundation model workloads to every domain.
Last updated September 2026.
In English, the two versions do not overlap by a single day. Registration for the MLA-C02 beta (exam code ME1-C02) opened on September 1, 2026. MLA-C01 can be taken in English until September 28, 2026, and MLA-C02 beta delivery begins on September 29, 2026. The beta is English only; the standard MLA-C02 exam is expected in all four exam languages in early 2027.
So the practical question is not "which exam is better" but "when is my test date". If you are booked and ready before the cutoff, sit MLA-C01. If you are not, every hour of study from now on should be pointed at MLA-C02.
| MLA-C02 | MLA-C01 | |
|---|---|---|
| Offered (English) | Beta from Sep 29, 2026 | Until Sep 28, 2026 |
| Scope | Traditional ML plus foundation models, GenAI, and agents | Traditional ML |
| Domain 1 | Data Preparation for ML and AI (28%) | Data Preparation for ML (28%) |
| Domain 2 | ML and Foundation Model Development (24%) | ML Model Development (26%) |
| Domain 3 | Deployment and Orchestration of ML and AI Workflows (24%) | Deployment and Orchestration of ML Workflows (22%) |
| Domain 4 | Operating, Monitoring, and Securing ML and AI Solutions (24%) | ML Solution Monitoring, Maintenance, and Security (24%) |
| Questions | 65 (50 scored, 15 unscored) | 65 (50 scored, 15 unscored) |
| Time limit | 130 minutes | 130 minutes |
| Passing score | 720 (100 to 1,000 scale) | 720 (100 to 1,000 scale) |
| Question types | Multiple choice, multiple response | Multiple choice, multiple response, ordering, matching, case study |
| Results | Delayed until the beta closes | Normal timing |
Dates, weights, and format from the official MLA-C02 exam guide and its comparison page, and the official MLA-C01 exam guide.
AWS did not add a fifth domain for generative AI. The exam guide's comparison page maps all twelve MLA-C01 task statements onto the twelve MLA-C02 tasks one to one, and the new material sits inside tasks you already knew. The weights moved by two points: Model Development dropped from 26% to 24% and Deployment and Orchestration rose from 22% to 24%.
Loading training data from EFS and FSx, model size reduction, edge optimization with SageMaker Neo, bring your own container, and two infrastructure monitoring statements. If you studied those, the time is not wasted, but they no longer earn marks.
The SageMaker AI spine is intact: data preparation with the AWS data stack, built-in algorithms and hyperparameter tuning, the four inference options (real-time, serverless, asynchronous, batch), SageMaker Pipelines and CI/CD, Model Monitor and drift, and least-privilege security. MLA-C02 also drops ordering, matching, and case-study questions, so the format is simpler even though the scope is wider.
You are already booked for a date on or before September 28, 2026, your practice scores are where you want them, and your work is classic ML on SageMaker. It is the version your preparation was built for.
You are not booked yet, your plan runs past September 28, or you work with Bedrock, RAG, or agents. Do not squeeze a ten-week plan into three days to beat the cutoff.
If you were ready for MLA-C01 and missed the window, the gap is smaller than it looks. Our MLA-C02 study plan estimates 3 to 5 weeks to close it, because the additions are concentrated in Bedrock, embeddings and RAG, agents, and guardrails. One trade-off to weigh: beta results arrive only after the beta closes, so if you need the credential on a fixed date, factor that delay in.
The last day to take MLA-C01 in English is September 28, 2026. From September 29, 2026 the English exam is the MLA-C02 beta (exam code ME1-C02).
MLA-C02 covers more ground, because it adds generative AI, agents, and foundation models to every domain, but the format is the same 65 questions in 130 minutes with a 720 passing score. It also drops the ordering, matching, and case-study question types, so the scope is wider while the format is simpler.
Mostly, yes. The SageMaker AI core carries over and every MLA-C01 task maps onto an MLA-C02 task, but MLA-C01 material does not cover Amazon Bedrock, RAG, agents, or guardrails, and a few topics such as SageMaker Neo were removed.
MLA-C02 launched as a beta, and AWS reports beta results after the beta closes rather than at the end of your session. The standard pass or fail designation does not apply during the beta.
No. The MLA-C02 beta is offered in English only. The standard MLA-C02 exam is expected in all four exam languages in early 2027.
Certified's MLA-C02 course covers all four domains in 76 short chapters, with a practice exam per unit and three full-length 65-question mocks.
Open the MLA-C02 course →