The December 2025 SAT: what students actually experienced

Hard, but not the hardest. Students put December below October and above every released practice form. That second half matters more, and you hear it after almost every real exam.

Inside the exam the shape was lopsided. Reading and Writing opened gently and stepped up sharply in Module 2, which several students found more disorienting than the difficulty itself. Math ran the other way: a manageable Module 1, then a Module 2 that either turned genuinely hard or didn’t, depending on where you were routed.

One thing was mostly missing from December feedback. Almost nobody reported running out of time. December wanted the right idea, not speed.

One December math question reached us more than any other, and almost never as a request to explain it. A circle given in general form with an unknown constant, paired with a second circle of twice the diameter. What students wanted to know was who else had seen it.

They had worked out something correct. That question only appeared on the harder Module 2. What they did with it is where it goes wrong. Students who saw it read it as bad news. Students who didn’t read it as relief. It is the reverse. The Digital SAT routes you into one of two versions of Module 2 based on how Module 1 went, and the harder path carries the higher score ceiling. Getting it means Module 1 went well enough to earn a form that can reach a high score. Not getting it means the ceiling came down early.

So the outcome is largely settled before Module 2 starts. The module you found comfortable is the one that settled it. Comparing questions with friends afterwards tells you about routing, not performance. If a friend describes questions you never saw, you have learned which form each of you got and nothing about how either of you did.

The other thing students asked was which items hadn’t counted. The same two came up over and over: that circle question, and a dense science passage late in Reading and Writing.

The premise is real. The SAT does embed unscored pretest questions to calibrate items for future forms. What students build on top of it isn’t. Notice which questions get picked out. Always the hardest ones. The exercise detects your discomfort, not a pretest item. A real pretest question has to be indistinguishable from a scored one, because an item that announced itself would be useless for calibration. So a question you couldn’t do almost certainly counted, and the two weeks you spend deciding otherwise buy you nothing.

December’s hard math wasn’t heavy on computation. It was built around stated conditions you translate into an equation before any calculating starts. A quadratic described as having no real solutions. A circle constrained by its relationship to a second circle’s diameter. A set of angle measures proposed as equivalent to a given one.

None of them asks what the value is. Each asks what has to be true. Take the quadratic. No real solutions means the discriminant is negative, so b squared minus 4ac is less than zero, and that inequality is the whole question. You solve it for the constant and stop. Nothing gets graphed and nothing gets factored.

The circle works the same way. General form hides the center and radius, so you complete the square to find them, and only then does the relationship to the second circle mean anything. Twice the diameter is twice the radius, which is one equation once you have the radius in hand.

This is also where Desmos stops helping. Students used it happily on a rational expression question, where the job is confirming two forms match. It does nothing for the discriminant question, because the hard part is knowing what to type. December’s Module 2 was put together so the calculator checks answers rather than finds them.

Vocabulary carried the section, and the reported words weren’t obscure. Students told us they knew these words and still couldn’t place them. That is what a usage question feels like from the inside.

The choices are plausible near-synonyms, so the sentence decides. It decides on connotation, and on which word the surrounding phrasing will actually accept. Take imbue. You imbue something with a quality, so the sentence has to supply both a thing and a quality going into it. A near-synonym like instill wants a person on the receiving end. You could define both and still pick wrong if you never checked what the sentence was offering.

Precarious and unwieldy split the same way. Precarious means it might fall, unwieldy means it is hard to handle. A heavy load is unwieldy. A load balanced on a ledge is precarious. If the sentence has been describing balance, only one of them is available to you.

A quieter theme ran under everything else. Forms weren’t all put together the same way, and students noticed. Some found one question type missing entirely, replaced by extra questions of another. Others hit long consecutive runs of a single type late in a module.

That unsettled people more than the difficulty did, and it carries a real lesson. Prep advice built on expected question counts, the kind promising a fixed number of items of a given type, is unreliable. What stays stable across forms is the set of skills and the shapes they arrive in. What varies is the mix.

Original questions written around what this exam tested, one for each argument above. Nothing here is a real test question.

The half of the verdict worth keeping is the second half. December ran harder than the practice forms, and so does almost every real exam. Practice comes from an archive of retired questions, while the live test keeps finding new ways to ask the same skills.

So more practice tests isn’t the useful preparation. Deliberate work on the small number of shapes that keep reappearing is, and December made those shapes obvious. Learn the six conditions above and you have most of its math.