Don't guess your level. Measure it.
Fifty-five to sixty-five minutes, five or six sections. At the end you get your CEFR level, the exam scores it corresponds to, and exactly where your accuracy starts to fall — from data, not from a hunch.
correct
TOEFL 4.0–4.5 · IELTS 5.5–6.5
What it measures
All four skills, in one sitting.
The speaking section is up to the institution that issued your code: when it is on, all four skills are measured and a total score estimate is given; when it is off, the report says so plainly and produces no total. We don't produce numbers for something we haven't tested.
- Listening10 questions
- Three clips: a campus conversation, a short lecture and an academic discussion. Each clip plays once, with one replay allowed.
- Grammar10 questions
- Structure and written expression: tense and aspect, articles and determiners, embedded question order, complementation, conditionals and relative clauses. Each item targets one documented point rather than testing grammar in general.
- Vocabulary8 questions
- Academic vocabulary, meaning from context, synonymy and word form. Nothing that rewards a memorised list — every item is solved from the logic of the sentence.
- Reading12 questions
- Three passages rising from B1 to C1. Main idea, detail, inference, pronoun reference and author's purpose are measured separately.
- Writing1 essay
- The current TOEFL academic discussion task. Scored against a fixed rubric in two independent passes; when the two disagree, a human reviews it.
- Speaking4 tasks
- An independent opinion task, read-then-speak, listen-then-speak, and an integrated task. Pronunciation is judged by a model that listens to the recording, fluency is measured from word-level timings, and language and content are rated against a fixed rubric in two passes.
How it works
Four steps, about an hour.
- 01
Enter your code
Codes are issued by a teacher or institution and belong to one student. You confirm your name before starting.
- 02
Check your sound and microphone
The exam opens with listening, so you play a short test clip first. On iPhone, audio plays silently with no error at all when the phone is on silent — this screen catches that. If the speaking section is on, you confirm your microphone is really recording with a record-and-listen check right before it.
- 03
About an hour
Each section is timed on its own, so you can't spend your reading time on grammar. The clock runs on the server.
- 04
Get your result
Your CEFR level, the exam scores it corresponds to, your difficulty curve, and an explanation for every question you got wrong.
Why the result holds up
The answer key never reaches your browser
Questions are sent without their answers. Scoring happens on the server.
Options are shuffled per attempt
"The answer to question seven is C" is worthless, because the order differs for every student.
Every code is tied to one student
The student confirms their name before starting. The result is saved under that name.
Your answers save as you go
If the browser crashes or your connection drops, you continue where you left off.
Sample report
This is what you get when the exam ends.
The report below came out of the real scoring engine. The numbers belong to a representative student; the calculation is genuine.
The shaded band is measurement uncertainty. No one-hour test can claim a single point; about one in four students near a cut score moves to the neighbouring band on a retest.
Estimated equivalents
- TOEFL iBT · Reading
- 4.0 – 4.5
- TOEFL iBT · Listening
- 4.0 – 4.5
- TOEFL iBT · Writing
- 3.0 – 4.0
- TOEFL legacy scale
- 72 – 94 reported until 2028
- IELTS
- 5.5 – 6.5
- PTE Academic
- 59 – 75
- Cambridge English Scale
- 160 – 179
Difficulty curve
You answered only a third of the C1 items correctly. Your ceiling looks to be around B2–C1 — that is where your study should begin.
Receptive and productive skill
reading and listening
writing
You understand at a higher level than you can produce. This is common and it is what surprises students most in the real exam: the writing section pulls the total score down.
Why you got them wrong
Question 14 · If the samples ______ stored correctly, the results would have been reliable.
You answered: wereCorrect: had been
The main clause has “would have been”, so this is a third conditional: something that did not happen in the past. The if-clause needs the past perfect. Learners routinely mix the second and third conditional because many languages build both with a single form.
For institutions
See your whole class on one screen.
For courses, schools and language centres: hand out codes, review results question by question, and know exactly where each student stalls.
Get in touchCodes in bulk
Generate as many codes as you have students and bind each one to a name. Each code is good for one exam.
The full answer sheet
Every question for every student: what they marked, what was correct, how long they spent, how many times they changed the answer.
Build your own exam
Write your own questions and texts and put them together into your own exam. On your own exam, results are given as a raw score.
Copy a Cefrly exam
Copy a Cefrly exam. The questions, texts and tasks become yours to change however you like.
Essay and speaking review
See the reasoning behind each score, read the essay and listen to the speaking recording. On an exam you built yourself, you can change a score you disagree with.
CSV export
All results in your own spreadsheet. The data stays yours.
FAQ
Questions, answered honestly.
Is this an official exam result?
No. This is a placement test: it tells you where to start. The TOEFL, IELTS, PTE and Cambridge figures are estimated ranges, not official scores. You cannot apply anywhere with this result.
Is speaking measured too?
Yes, but it is optional: the institution that issues your code decides whether the speaking section is on. When it is, you get four tasks; your recordings are transcribed, fluency is measured from the audio itself, and pronunciation is rated by a separate model that listens to the recording. If the two ratings disagree, it goes to human review. If you have no microphone you can skip the section and finish the test, then sign in later with the same code to complete just the speaking part — that second sign-in does not use up an attempt.
How reliable is the result?
On a well-built 40-item test the standard error of measurement is about three items. In practice: if you land close to a level boundary, roughly one in four students moves to the neighbouring band on a retest. That is normal, and the report shows it as a confidence label rather than hiding it.
Can the same student take it again?
Each code is good for one exam. If a student needs to sit it again, the institution gives them a new code. A student without a microphone can finish the exam and complete the speaking part later with the same code; that does not count as a second attempt.
Where do the questions come from?
All of them are ours and written by hand. Reading passages are derived from copyright-free sources (Wikipedia, NASA, NOAA, open-access journals) and their difficulty is verified by measuring word frequency and sentence length. Not a single item is taken from an exam-preparation site.
What happens to my data?
Your name, answers, essay and any audio recording are stored on a server in Frankfurt. Your essay and speech are sent to AI services for scoring. Audio is kept for at most 365 days, your name and free text for 24 months; scores and answers then stay without a name. Details are in the privacy notice.
Start
Have a code? Enter it here.
Eight characters; you do not need to type the CRF at the front. Before you begin you will confirm your name and pass a short sound check.
No code?
Codes are distributed by teachers and institutions. If you're enrolled in a course, ask your instructor. For institutional use, write to us.
info@apperable.com