Cost breakdown

PMLE

How much does Professional Machine Learning Engineer cost?

Professional Machine Learning Engineer is $200, and the lab bill is the one to watch: training jobs and deployed endpoints on Vertex AI bill by the hour, and an endpoint left up over a weekend costs more than most people expect. Inside the $300 trial credit and with endpoints deleted after each session, the practical cost is the exam fee. Budget $200 self-studied, up to $700 with paid training and a retake reserve.

Cost breakdown

ItemCostDetail
Exam voucher$200The published US list price for one attempt at PMLE.
Retake$200A second attempt costs the same as the first.
Renewal examVariesGoogle publishes no shorter renewal exam for its professional certifications. Renewing means the full exam at the full price.
Google Cloud account for hands-on practice$0The free tier plus $300 in signup credits covers what you need. Set a budget alert on day one and delete what you build.
Google Skills learning pathVariesGoogle publishes a learning path for each exam. Some of it is free and some sits behind a subscription, so check the price before you buy anything.
Exam guide$0Six sections with weightings. Scaling prototypes is 21 percent, serving and scaling 20 percent and pipelines 18 percent, so nearly 60 percent of the exam is about getting a model into production rather than about building one.
Free trial credit$300, for 90 daysEnough for this curriculum if training runs are small and endpoints are short-lived. The exam does not ask you to train anything large; it asks which hardware and which serving pattern fit a described constraint.
Vertex AI endpointsHourly while deployedThe most common source of unexpected spend on this exam. An endpoint bills for the node behind it whether or not a prediction is served, so undeploy the model at the end of every session.
BigQuery MLFree tier, then per byteThe low-code section is 13 percent of the exam and BigQuery ML is the cheapest way to practice it, because training happens inside a query rather than on a dedicated cluster.
Paid learning path$0 to $400Optional. Most useful for MLOps and pipeline material, which is the part practicing engineers most often have never had to build end to end.

Exam pricing from Google Cloud, Professional Machine Learning Engineer exam details ↗.

Ways to bring the cost down

Undeploy every endpoint before you close the laptop. This single habit is the difference between a $30 lab and a $300 one, because endpoint nodes bill continuously and silently.

Train small. The exam tests judgement about hardware choice, distribution strategy and cost, and those decisions are learned from the documentation and one small experiment rather than from an expensive run.

Use BigQuery ML for the low-code section. It costs almost nothing at study scale and covers a section worth 13 percent.

Set a budget alert before the first training job. Machine learning services are the easiest place on Google Cloud to spend real money by accident, and the alert takes two minutes.

  • 50 percent renewal code

    Passing puts a 50 percent discount code in the Benefits section of your CM Connect account, and you can apply it to your renewal attempt. Failing does not earn one.

    Google Cloud, vouchers and discounts ↗

  • Free tier and trial credits

    Over 20 Google Cloud products have a free monthly usage tier, and new customers get $300 in credits on signup. That covers most of the hands-on work these exams ask about if you delete what you build.

    Google Cloud free tier ↗

For a good share of candidates somebody else pays this fee. The GI Bill, DoD and employer routes, with the rules each one publishes.

What a retake costs

A retake is another $200 at full price.

Google's ladder applies: 14 days after a first failure, 60 after a second, 365 after a third, four attempts in any two-year period.

No published passing score and no section breakdown, so the result is pass or fail with no diagnosis attached.

Failures here usually come from the operational sections rather than the modelling ones. Data scientists who can build a model lose marks on pipelines, retraining triggers and monitoring for skew, which is 31 percent of the exam between pipelines and monitoring. Between attempts, build one pipeline that retrains on a schedule and alerts on drift, even if the model itself is trivial.

Retake pricing from Google Cloud, Professional Machine Learning Engineer exam details ↗.

What renewal costs across the full cycle

This is a Professional certification, valid for two years, with the renewal window opening 60 days before expiry.

Renewal means passing the applicable exam in that window, and exam fees apply. Shorter renewal exams and a Google Skills learning route exist for certifications that support them.

Two years is short and on this subject it is barely enough. The Vertex AI surface and the generative material inside this curriculum have moved faster than any other area Google certifies, so a renewal here is real study rather than revision.

The post-pass discount code applies to a renewal attempt, which is worth planning for rather than discovering after you have paid full price.

Typical preparation budget

Preparation for ML Engineer can genuinely cost nothing. The realistic range people spend is $0 to $400, and the top of that range buys convenience rather than a better chance of passing.

Our questions, explanations, and cheat sheets are free, and the explanation is never paywalled. That is the part of a paid practice set you are actually buying, so it is the part we do not charge for.

The cheapest way to avoid the $200 line on this page is to be ready before you book. Free ML Engineer practice questions with a full explanation on every answer.

Keep reading

Every guide and cost breakdown, by vendor

Practice ML Engineer for free while you decide

Original questions written from the published objectives, with the concept, the reasoning, and a note on every wrong option. No account needed to start.

Start free ML Engineer questions