The June 2026 SAT: the hardest math of the year, and the calculator stuttered

Yes, and the math above all. June drew roughly four times as many difficulty reports as reassurances, the most lopsided reaction of any recent test date, and the phrase students reached for repeatedly was that this was the hardest math they had seen.

A second theme ran alongside it, and it is the only sitting where this came up at volume: students describing the graphing calculator as slow, stalling, or unresponsive at the moment they needed it. We cannot verify how widespread that was and would not try. What matters is the effect it had on the people it happened to, which was to remove, mid-module, the method they had built their preparation around.

Reading and Writing was quieter but not soft, and vocabulary carried it. The words students brought back from June make up the widest and most demanding list of any recent form.

Desmos is a genuine advantage on the Digital SAT, and any serious preparation should include learning it properly. The problem is the way students end up using it, which is not as an accelerator but as a substitute for a decision.

You can see the difference in how a question gets approached. A student with a method reads a tangent-line question, recognises that the radius meets the tangent at a right angle, writes the slope relationship, and then may well use the calculator to finish the arithmetic quickly. A student without one opens the calculator first, plots what they have been given, and looks at it hoping the answer will become visible. On easy items the second approach works often enough to feel like a technique. On hard ones it produces nothing, because there was never anything to plot.

June is what that looks like under pressure. Even setting aside the reported stalling, the section was built around questions where the work happens before the calculator can be involved: a data set described only by its median, range and mean; a circle described by a point of tangency and a slope; a growth rate expressed over a fractional interval. None of those can be typed in until you have already made the key decision.

The practical test of whether you have a method is simple. Work a set of hard math questions with the calculator closed, allowing yourself to open it only once you have written the equation you intend to solve. If you cannot get started on most of them, the calculator has been carrying more of your preparation than you realised, and June is the form that would have exposed it.

The specific shape that made June’s math hard deserves naming, because it recurs and it is trainable. Several items described a data set entirely through its summary statistics and asked for one of its values. You are told the mean, the median and the range of five integers, three of which are known, and asked for the largest.

There is no reading anything off a graph here, and no formula to apply directly. The route is to convert each summary statistic into an equation about the unknowns: a mean gives you a total, a range gives you a difference between the extremes, a median fixes the position of the middle value. Three constraints, two unknowns, and the answer falls out.

This is the same reversal that has been running through every recent sitting, arriving in the statistics domain. The question describes properties, and your job is to write the conditions those properties impose before solving anything.

June’s hard math clustered in three places. Circles, and specifically tangency, where the whole question is the right angle between radius and tangent. Statistics described through summary values rather than listed data. And exponential growth stated over a fractional interval, where the exponent has to absorb the interval before the rate makes sense.

Each is the same underlying task in different clothing: read a described property, convert it into an equation, then compute. The computation in every case is small. Students who found June impossible were almost never stuck on arithmetic.

June’s word list is the broadest of any recent sitting, and it is a different register from the exam’s usual fare. These are words that appear in argument rather than description, and several of them carry a judgement: to call something spurious or inane is to evaluate it, not to categorise it.

That has a bearing on how the items work. When the options carry evaluative weight, the sentence usually signals which direction the writer’s judgement runs before the blank arrives, and the wrong answers are frequently right in meaning but wrong in attitude. The other pressure point was inference questions asking what a set of results undermines, which reward tracking a hypothesis through to what the evidence actually contradicts.

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

Whatever happened with the calculator, the lesson generalises. A technique that only works while a tool is available is not a technique, and June was built so that the tool could not be the first step even when it worked perfectly.

The specific correction is to practise the translation step separately. Take hard questions and write only the equation you would solve, without solving it. That is the part June tested, it is the part Desmos cannot do for you, and it is the part that stays useful on every form that follows.