Exam Prep

Verified against ChatGPT · 2026-08-12

Build a full-length timed mock test that simulates real exam pressure, with answers held back until you actually finish

Assembles a complete, section-timed mock exam matching your real exam's structure and difficulty, deliberately withholding the answer key in a separate follow-up so you take it cold under realistic timing.

ChatGPT (GPT-5.1)4 fillable variables

The prompt

Ready to copy — highlighted parts are example details you can swap.

Build me a full-length mock test that mirrors the real exam's structure, section timing, and difficulty as closely as possible, and hold back the answer key entirely — do not include it in this response.

EXAM BEING SIMULATED
GRE General Test

SECTION STRUCTURE AND TIMING
Two Verbal Reasoning sections (20 questions, 30 min each), two Quantitative Reasoning sections (20 questions, 35 min each), in that order

TOPICS TO WEIGHT TOWARD
Quant sections should be roughly 40% algebra, 30% arithmetic/data interpretation, 30% geometry, matching the real test's typical mix

DIFFICULTY CALIBRATION
Calibrate to the difficulty of official ETS practice test 2, not an easier introductory level

PHASE 1 — BUILD THE TEST
Generate original questions for every section in the structure given, in the same order and with the same per-section time allocation as the real exam, so the whole thing can be printed or opened in one document and attempted start to finish with a single timer. Match difficulty to the reference given, not to what feels achievable — a mock test that's easier than the real exam produces false confidence, which is worse than no mock test at all. Distribute topic coverage according to the weighting given, not evenly across everything you can think of. At the top, state the total time allowed and each section's individual time budget clearly enough that I can set timers before starting.

PHASE 2 — WITHHOLD ANSWERS
Do not include any answers, mark schemes, or worked solutions in this response, even partially. End this response with only the line: "Reply 'grade it' once you've finished and timed yourself, and I'll provide the full answer key and a section-by-section score breakdown." This is a hard rule — a mock test with the answer key visible in the same response isn't a mock test, since it can't be attempted honestly under pressure.

OUTPUT FORMAT
Title and total/section timing block, then each section's questions in full, ending with the single withholding line above and nothing else.

Customize

Optional — swap in your own details for the highlighted parts above.

Why this works

GPT-5.1's helpfulness training pushes it toward completeness by default, which for a request like "build me a mock test" usually means appending the answer key immediately after the questions so the response feels self-contained and useful in one shot — the explicit, repeated, hard-rule instruction to withhold answers and end on a single specific line is necessary because a softer phrasing like "don't show me the answers yet" is easy for the model to satisfy nominally while still leaking partial hints in surrounding commentary. Structuring the request as two phases, with Phase 2 defined entirely as a withholding action rather than a generation action, gives the model an explicit stopping point to execute rather than leaving "how much to hold back" as a judgment call it has to make mid-generation, which is where hint-leakage usually creeps in. The difficulty-calibration field matters because GPT-5.1, left to its own judgment about exam difficulty, tends to regress toward a moderate, broadly-accessible level that undershoots genuinely hard real exams — anchoring difficulty to a named reference (a specific official practice test) gives it a concrete target to match rather than an internal, uncalibrated sense of "hard enough," which is exactly the failure mode that produces a mock test easier than the real thing and the false confidence that comes with it.

What you get back

GRE Mock Test — Total time: 130 minutes. Section 1: Verbal Reasoning (30 min, 20 questions)... Section 4: Quantitative Reasoning (35 min, 20 questions)... [full questions follow for each section] ... Reply 'grade it' once you've finished and timed yourself, and I'll provide the full answer key and a section-by-section score breakdown.

Verified against

ChatGPT GPT-5.1 · 2026-08-12

Changelog

  • 2026-08-12 Initial publish, verified against ChatGPT GPT-5.1.

Need this built into your business?

If a prompt isn't enough — what Scult builds, built and maintained for you — that's Scult's day job.

EXPLORE WHAT SCULT BUILDS
All Exam Prep prompts

Check your AI visibility

One URL in, a 0–100 score and the exact fixes out.

RUN THE CHECK

Browse all the tools

15 tools across six categories
13 of them never send your data anywhere

Free · No signup · No trial clock

SEE THE DIRECTORY