Full simulation · free · no sign-up
Timed mock exam
A full-length practice exam under real conditions: 63 questions in 120 minutes, drawn in the same domain proportions as the published blueprint, with nothing revealed until you submit. It is scored on the same 100–1000 scale the real exam uses, and the report tells you how far you landed from the 720 cut.
Loading…
What the CCAR-P mock exam covers
The form is drawn in the proportions the published blueprint gives each domain, so the balance of what you sit matches the balance of what the real exam asks. Within a domain the draw is stratified by objective: every objective contributes an item before any objective contributes a second. Plain random sampling is what used to let whole objectives go unexamined across a full 63-item form.
| Domain | Blueprint weight | Items drawn |
|---|---|---|
| D1 · Solution Design & Architecture | 17% | 11 |
| D2 · Models, Prompting & Context Engineering | 13% | 8 |
| D3 · Integration | 19% | 12 |
| D4 · Evaluation, Testing & Optimization | 16% | 10 |
| D5 · Governance, Safety & Risk Management | 14% | 9 |
| D6 · Stakeholder Communication & Lifecycle | 14% | 9 |
| D7 · Developer Productivity & Enablement | 7% | 4 |
How the mock is scored
Scoring follows the real exam's rules rather than a friendlier version of them. Your raw percentage is converted onto the same 100 to 1000 scale, and the report shows the margin between where you landed and the 720 cut.
- Single-answer items are marked on the exact option. Nothing is given for being close.
- Multiple-response items need the whole set and only that set. There is no partial credit for finding two of the three.
- Matching items are all-or-nothing across every pair in the item.
- Anything left unanswered counts as wrong, which is why the clock is worth watching. A guess and a blank score the same when the guess is wrong, and better when it is not.
Where this differs from the real exam
Worth being straight about the limits. These questions are original, written against the published blueprint; none of them are real exam content, and a practice set that claimed otherwise would be telling you it had broken an exam agreement. Scoring well here is evidence that you know the objectives, not a prediction with a confidence interval on it.
The sitting conditions differ too. The real CCAR-P is proctored through Pearson VUE with the checks that implies, and it costs $175 against a registration that runs through an Anthropic partner rather than a personal email address. This runs in a browser tab you can close. The part that transfers is the question shape and the objective coverage, and that is the part worth practising.
How long should I spend per question?
120 minutes across 63 items is a shade under two minutes each. Most items take well under that and a few scenario items take considerably more, so the useful habit is flagging anything that runs long and coming back rather than spending six minutes proving a point to yourself.
Does the clock keep running if I close the tab?
Yes. The countdown runs against the wall clock, not against time spent on the page, so closing the tab does not pause it. The sitting is saved and can be resumed, but it will come back with the time that actually elapsed already deducted.
Can I sit the mock more than once?
As many times as you like. The form is redrawn from the 200-item pool each time and the options are reshuffled, so a repeat sitting is not the same paper. Repeats do get easier for the wrong reason once you have seen most of the pool, which is worth remembering before reading too much into a rising score.
Sitting a mock cold is a diagnostic. The guide walks all 38 objectives in the blueprint, which is the part you study between attempts.
Read the study guide