Off by default, deliberately. Full mocks stay on the real exam's difficulty mix so their score stays comparable to a real SAT — turn this on and mock scores will read lower than the real thing.

Matches the College Board accommodations. Only set this if she tests with the same accommodation — practising with more time than she gets on the day teaches the wrong pacing.

Once you have logged a Bluebook score, practice estimates are shifted to agree with College Board where you have both. This is the closest a third party can get to real equating — capped at 80 points so one odd test can’t distort the scale.

A miss is only worth what you learn from it. Before the key and the explanation appear, she has to say why she got it wrong. Easy items are usually careless rather than conceptual, so gating them adds friction without teaching much — that is why medium-and-hard is the default.

Picking teaches the vocabulary test writers design against, and it is fast enough to do on all 20 misses. Typing is harder and more honest, but it is graded by AI and slows a long review down.

Choose any enabled provider — Claude, GPT, or Gemini. The list comes from the server; add or change models in api/config.php.

Strongly recommended for accuracy. Applies to live generation — bank questions are already verified before they're stored.

📦 Question bank: checking…

Tests pull instantly from a bank of pre-written, verified questions. It refills on a schedule (cPanel cron) and quietly after each test. Top up now

Today's report sends at this time if the app is open then, or on the next app open after it. Practice done after it's sent shows in lifetime stats but not in that day's report. Clear the field to only send finished days.