Free AI scoring for PTE Academic
Practice under real exam timing. Get a 10–90 score in four seconds, with the exact word that cost you points.
Free forever · No card required

507
questions in the bank
22
task types — every one Pearson sets
4
skills scored on the 10–90 scale
~4 s
from “stop recording” to a score
Scored like Pearson
Speaking answers are recorded at 16 kHz, transcribed with word timings, and scored on content, oral fluency and pronunciation as separate traits — never averaged into a vibe. Hesitation becomes a measured gap; a stumbled word gets named.

Speaking
Transcribed with word timings; content, fluency and pronunciation scored separately.
Writing
Form checked before content — the order the exam applies.
Reading
Blank-by-blank partial credit on every fill-in and reorder task.
Listening
One-play audio, word-by-word dictation marking.
The product
Practice, mocks, revision and vocab write to the same record, so the dashboard always knows what to hand you next.

Practice average, weekly target, mock trend and a three-step plan built from your weakest task types — not a wall of stats.

Three parts, about 125 minutes, exam-weighted and drawn fresh from the bank. Or drill one skill on its real section clock.

Every task type has its own bank with difficulty, frequency and practised/new filters, so you stop scrolling and start answering.

Dictation words, fill-in-the-blank answers and a core list — drilled in reading or listening mode until they stop being hard.
How it works
Start with the one you dread. All 22 are there, including Summarize Group Discussion and Respond to a Situation.
Preparation and response clocks run exactly as the test runs them. Audio plays once. Recording starts itself.
A 10–90 score, the trait that dragged it down, and the sentence or word that cost you. Then the question comes back on a spaced schedule.
Straight answers
No — nobody outside Pearson has their engine. Our scores are estimates built from the published rubrics and scaled to 10–90. Use them to find weak spots and track direction, not to predict your report card to the point.
Yes. Summarize Group Discussion and Respond to a Situation are in practice, in the full mock, and weighted the way the exam weights them.
An account, the practice library, trait-level scoring and the revision queue. No card, no trial countdown.
Yes. The institute portal adds class rosters, join codes, assignments and submission marking on top of the same scoring engine.
From the transcript’s word-level confidence and timing. It is the hardest trait to automate for anyone, so read a low score there as a prompt to listen back, not a verdict.
Make an account, pick the task type you dread most, and read what the breakdown says. No card, no sales call.