Your topics now decide what you practise. Sessions used to serve whichever track held the day's focus skill and nothing else, so Algebra and Calculus could sit selected and unpractised. Every topic you are studying now gets a share of every session, and Today tells you how many questions each will get. Choose your topics at the start, and change them whenever you like in Settings. Placement only asks about what you chose, so a narrower selection is a shorter start. Progress opens up. Number Fluency breaks into its five areas — addition and subtraction, multiplication and division, fractions, percentages and ratios, estimation — each with its own level, instead of one number for all of it. A level you have not earned yet says so rather than showing a figure you did nothing for. Sessions stop repeating themselves. A topic with a small pool of questions no longer asks the same one several times in a row. A finished topic says what it is: every available skill mastered, and no review due today.

Estimation was hardest in the middle of the ladder rather than at the top, which nobody intended. The accepted range now eases more gently as challenge levels rise, so the jump around the middle is no longer the steepest part of the climb. You can now pause a skill. A paused skill stops appearing in daily practice and keeps everything it has already unlocked, so setting one aside costs you nothing. Reckon now measures how answers land, so unfairly hard questions can be found and fixed. What is sent carries nothing that identifies you. Data & privacy explains it and lets you decide whether yours are grouped together.

Estimation now rewards a sensible range instead of one exact number, including percentage questions. Every missed question now gets a relevant worked correction using the numbers you actually saw. Practice, hints and due reviews also update progress more faithfully, with clearer challenge-level and method language throughout.

When an answer goes wrong, the working is now done on your numbers. The method is shown step by step on the sum you actually saw, rather than on a stand-in example. Where a method genuinely does not fit the numbers you were given, the general worked example is shown as itself, instead of being bent to fit. Smaller things: tertiary text, control edges and unavailable labels now meet contrast in every theme, so nothing that carries meaning is left faint. And if your progress cannot be saved, the opening placement lets you in and says so plainly, rather than asking you to try again.

Sessions can now be six questions, for the days when twelve is too many. Your practice has a history of its own: every session you have finished, grouped by week, with any level that moved. The methods you have unlocked are readable again, in one place, whenever you want them rather than only when a question goes wrong. On iPad, every screen now uses the width — a side rail, two columns where they earn their place, and a list beside its detail on Settings, Progress and your history. Widgets are down to the ones that carry something, and say so when they are showing yesterday. The reminder now asks once, after your first finished session, rather than before you have anything to be reminded about. Smaller things: a correct answer no longer says the same sentence every time, and your levels are read the same way on every screen that shows them.