What we build against, and how to check us.
Vettamed makes one promise: nothing on exam day should be new to you. That promise is only worth something if you can verify it — and normally you could not. There are no released HUMAT past papers to measure us against. We publish no testimonials, because we have not earned any. We publish no pass rates, because we have not measured any. So this page is what we offer instead: the exact shape we build to, the choices we made and why, what we refuse to claim, and what is still missing today.
The shape every paper is built to
The 2026 HUMAT runs as four separately timed phases. Each is submitted on its own, and once submitted it is closed to you. 66 questions. 132 minutes.
| Phase | Internal split | Questions | Time |
|---|---|---|---|
| Scientific Thinking & Academic Literacy | 10 then 10 | 20 | 45 min |
| Biology & Chemistry | 10 then 10 | 20 | 35 min |
| Mathematics & Physics | 10 then 10 | 20 | 40 min |
| Reserve | one per subject | 6 | 12 min |
That is the shape every Vettamed paper is built to: the same phase order, the same internal subject split, the same clock on each phase, the same close on submission. Five options per question, A to E, plus an explicit omission that scores zero — never multi-select, never free text, never "select all that apply". Papers are marked with the published weighting and the per-section floors, which the scoring guide sets out in full. We do not have full papers on sale today — see the last section for why.
Read how the HUMAT is scoredFour choices that look like gaps until you know why
Fidelity means matching the exam where it is generous and where it is not. These four are the ones most often got wrong, and each of them is checkable against the official material.
The official 2026 Chemistry section is entirely text-based. A product that adds diagrams there is not being generous — it is rehearsing you for an exam that does not exist. Our one figure-based chemistry topic, molecular geometry, is kept in practice only, deliberately.
The same rule read the other way. The official material supports them, so we build them. Fidelity is not minimalism; it is matching. Taking something away that the exam actually does is the same defect as adding something it does not.
The six reserve items exist to replace a question that gets neutralised. They are built to the standard of a scoring question rather than as easy filler, because on the day they may be the ones that count.
You cannot go back to check your working once a phase is closed. This is the property most preparation tools quietly drop, because it makes people unhappy while they study — and it is the single largest source of surprise on the day.
How a question earns its place
We do not write questions and then proofread them. Practice questions are produced by 91 generator families, and every family carries a solver that works the question out independently of how it was built. If the solver and the answer key disagree, the question does not enter the bank. The answer is not a claim somebody typed — it is a result that was re-derived.
For Scientific Thinking, that generation runs offline on our own machines with no language model anywhere in the path. The reasoning is computed, which is also why every one of those questions can explain itself: the explanation is the solver’s working written down, not prose composed about it afterwards.
There are 6,000 practice questions live today across 100 sets, and you are never served one you have already seen. Exposure is recorded permanently for each student, and a set that would repeat a question is not offered to you. That guarantee is the reason the bank is built the way it is rather than as a pile of files.
What we will not tell you
Each of these is something we could say today, and would sell more by saying. They are absent deliberately. The list is here so you can hold us to it.
- We will not predict your rank or your final score. Nobody can do that honestly without a model validated against real, verified outcomes, and we do not have one. When we do, we will say how it was built and show how wrong it can be.
- We publish no testimonials, no pass rates, and no success statistics. Not because ours would be unflattering — because we have not earned any yet, and the first ones will be real.
- We claim no affiliation with, endorsement by, or partnership with Humanitas University. Vettamed is independent. HUMAT and Humanitas are their names, used here to say plainly which exam this is for.
- We do not use your camera, your microphone, or your screen. There is no proctoring, no monitoring, and no plan to add any.
What is missing today
We had seven full mock papers on sale. On 27 July 2026 we withdrew all of them. They had been migrated from an older system, and this project’s own exam specification classifies that system’s output as what not to do — so every one of those 462 questions carried an unset verification flag, two of them were the same question, and five had answer keys that two independent attempts disagreed with. They were not good enough to charge for and they are gone. What is live is the practice bank above. Replacement papers are being built by the generator described here, and we will say so plainly when they exist rather than counting the old ones in the meantime.