The question bank is for studying a domain at a time, with the answer and the reasoning in front of you. This page is for the opposite: taking an attempt with nothing to lean on, and finding out where you actually are.
Questions are drawn at random and reshuffled on every attempt, as are the options within each question, so a second run is not the same exam with the answers in the same places. Miss one and the review will tell you why the right answer is right, and link you to the GitHub or Microsoft documentation that says so.
Exam simulator
117 questions drawn from the practice bank, shuffled on every attempt. Scored 0–1000 the way the real exam reports, with 700 to pass.
Nothing is sent anywhere and nothing is stored. Reloading the page ends the attempt.
The real exam uses a scaled score, so a percentage does not map to it exactly. This is a proportional approximation and should be read as a guide, not a prediction.
Every answer was checked against the documentation#
These questions came from a third-party practice bank, and those are not reliable. Rather than republish them on trust, every question here was verified against current GitHub Copilot documentation and Microsoft Learn.
That mattered. Five answers were simply wrong and have been corrected. Three more questions could not be answered cleanly as written, so they were rewritten against the current documentation so that each has one defensible answer. Three questions were removed entirely: two were duplicates, and one asked about the CLI while keying the IDE workflow.
Every explanation in the review reflects the documentation as it stands today, with a link to the page that backs it up.
How the scoring works#
Microsoft reports GH-300 on a scaled score from 0 to 1000, and 700 passes. A scaled score is not a percentage: the real exam weights items by difficulty, so a given number of correct answers does not map cleanly onto a fixed mark.
That weighting is not published, so this simulator maps your result proportionally onto the same range. It is a reasonable guide to whether you are ready, and it is deliberately not presented as a prediction of your real score.
Two details are matched to the real exam:
- Multi-answer questions are all-or-nothing. Selecting two of three correct options earns nothing, exactly as it would on exam day.
- Unanswered questions are marked wrong. There is no penalty for guessing on the real exam either, so answering every question is always the right move.
Practicing under time pressure#
The timer is off by default. Tick Add a 100-minute timer before you start and a countdown appears beside the question counter. It turns amber with ten minutes left and red in the final minute, and when it reaches zero the exam submits itself with whatever you have answered, just as the real one ends. The result then shows how long you took.
Leave it off while you are still learning the material, and switch it on once you want to know whether you can finish in time.
Before you take the real exam#
A practice bank is a diagnostic, not a syllabus. Passing this comfortably means you have learned these questions, which is not the same as knowing the material. The real exam draws from the whole outline and is refreshed regularly.
When something here surprises you, go and read the reasoning rather than memorizing the letter:
- The six exam domains for where the underlying material lives
- Question bank for the same style of question with explanations and source citations
- Hands-on labs for the parts that only stick once you have done them
- Plans and billing, consistently the most-missed area and the easiest to fix
The official practice assessment is free, unlimited, and weighted like the real thing. Take it cold before you study anything. Failing it comprehensively is the most useful twenty minutes you will spend, because it tells you exactly where the holes are.
