Skip to content
celpip9
Start free
Log in
A 30-second animation about celpip9. The four skills of the CELPIP® General test. Exam mode: “As the format is published. Nothing softened.” — the clock stays on screen and a Speaking task moves from preparation to recording on its own. An estimated level shown as a range, CLB 9 (8–10), invented for the example: “Every estimate comes with a range.” Feedback that revises the learner’s own sentences. The celpip9 name: “Practice you can check.” First full test free.

CELPIP® General practice · first full test free

Practise all 20 CELPIP® tasks. See your estimated level before test day.

Full practice tests and short task sets, built on the published format. Listening and Reading are marked against the answer key. Writing and Speaking get an AI estimate, always shown as a range.

Start free, no cardSee all 20 tasks
  • The timer never pauses
  • Every accuracy figure published with its sample size

What’s inside

  • Listening96 sets6 parts · 38 scored questions · marked against the answer key
  • Reading136 sets4 parts · 38 scored questions · marked against the answer key
  • Writing68 sets2 tasks · AI estimate, shown as a range
  • Speaking272 sets8 tasks · AI estimate, shown as a range
  • 16full tests
  • 572task sets
  • 20strategy lessons
  • 10vocabulary topics

Counted from the content bank when this page was built. Source · parts and questions: published test format

Free and paid

What you get, before and after you pay.

Free no card

  • 1 full test with AI feedback (Test 14 · Wrenfield, everyday topics), with its full report
  • Task sets: 1 set of each of the 20 parts and tasks (20 sets)
  • All 20 strategy lessons and 10 vocabulary topics, free
  • The test drive: every mechanic once, untimed and unscored
Start free

Monthly or 3 months $29 per month · $59 for 3 months

  • 2 full tests a week (about 8–9 a month)
  • All 572 task sets, with Writing and Speaking estimates
  • Whole sections: Any section of any of the 16 full tests, on its own
  • Progress over time, and the one thing to fix next
See prices and every limit

Every part of the test

4 skills, 20 tasks, a set of each from every test.

Task names as the test publishes them. A task set is one part or task from one of our full tests, on its own clock.

Listening

46–55 min section

Audio plays once. Questions appear when it ends.

Reading

43–56 min section

Passage and questions side by side; on a phone, one tap switches.

Writing

53 min section

Feedback quotes your own sentences and shows them revised. No templates.

Speaking

15 min section

Recording starts by itself when preparation ends.

SourceCELPIP-General test format (celpip.ca)

*Speaking preparation and answer times per task are a third-party breakdown; the test provider does not publish them. Word ranges are those of our banked tasks.

Every task with its set titles and topics

Your report

Estimated levels as ranges, then the one thing to fix.

Sample report · illustrative

Invented example

Estimated levels

ListeningCLB 9 (8–10)
ReadingCLB 6 (5–7)
WritingBeing scored
SpeakingBeing scored

Published 2024 mean for the skillCLB 9 and above

Source CELPIP Data Report of 2024 Test Takers (Paragon Research Reports, PDF)

Estimated from your number of correct answers with the approximate chart the test provider publishes. Our practice tests are not yet equated to the score scale the test itself uses, so read it as a range. Not a CELPIP score. Writing and Speaking: Estimated level, based on the published CELPIP Performance Standards dimensions. Our scoring is not yet equated to the score scale the test itself uses, so read it as a range. Not a CELPIP score.

Holding your level down

Missed paraphrase

The report picks the one pattern to fix first, with a retry aimed at it.

  • Ranges, never a single number

    Listening and Reading ranges come from the test provider’s approximate raw-score chart. Writing and Speaking ranges come from several independent scoring passes.

  • Your sentence, revised

    Writing and Speaking feedback quotes what you wrote or said and shows it revised, next to the dimension it affects.

  • Compared with the published results

    Each level sits beside the published 2024 results, so you see how far CLB 9 is and how many test takers reach it.

See a Writing and a Speaking report, task by task

Reached CLB 9 or higher, 2024

  • Listening52.0%
  • Reading33.3%
  • Speaking32.6%
  • Writing29.8%

Reading has the lowest mean level of the four: 7.21. Listening 8.22 · Writing 7.91 · Speaking 7.76.

SourceCELPIP Data Report of 2024 Test Takers (Paragon Research Reports, PDF)

How it works

Three steps, then repeat.

  1. Take a task set or a full test

    A short set for one task, or the full test in one sitting. Both run on a clock that never pauses.

  2. See your range and the one thing to fix

    Levels as ranges, every miss explained, and the single pattern that costs you the most.

  3. Practise that task with a new set

    A set you have not seen, the same task type. Then check the change.

How accurate we are

Every figure with its sample size. Or “not yet measured”.

Scoring consistency (the estimate you see, same response scored repeatedly)

0.13CLB

n = 30target < 0.5 CLBmeasured 2026-09-24

Raw pass spread (single scoring passes, same response scored repeatedly)

0.35CLB

n = 30target < 0.5 CLBmeasured 2026-09-24

Answer-key accuracy

100%

n = 1092target ≈ 100% (independent solver agrees with the key)measured 2026-09-30

Practice difficulty vs. published 2024 levels

Not yet measured

Target: deviation within ±0.25 level. Published when the measurement exists.

Reported and corrected items

Not yet measured

Target: published monthly. Published when the measurement exists.

Estimate vs. reported result

Not yet measured

Target: MAE ≤ 1 CLB. Published when the measurement exists. It needs test takers who share their result — you can be one.

Sourcecelpip9 evaluation pipeline output (packages/eval)

Full trust panel and how each metric is computed

Exam mode

Timed to the published format. Nothing softened.

When a section starts, the product steps aside: a neutral screen and a fixed bar with section, part and time.

  • One play per recording

    No replay, no seeking. The questions and their options appear on screen when the recording ends.

  • No pause, in any mode

    Once “Begin section” starts the clock, it runs to the end.

  • Speaking runs itself

    Preparation time, then recording starts and stops on its own, then it is saved.

  • Device check first

    Sound, and the microphone when Speaking is included, are checked before any clock starts.

Pricing · CAD

Quotas printed, nothing hidden.

For scale: one attempt at the test costs $295 + tax.

Free

$0 CAD · no card

Full tests
1 full test with AI feedback (Test 14 · Wrenfield, everyday topics)
Task sets
1 set of each of the 20 parts and tasks (20 sets)
Start free

Monthly

$29 CAD · per month

Full tests
2 full tests a week (about 8–9 a month)
Task sets
All 572 task sets
Choose monthly

3 months

$59 CAD · 3 months

Full tests
2 full tests a week (about 26 over 3 months)
Task sets
All 572 task sets
Choose 3 months

Only need tests? 3-test pack $19 CAD · one-time · valid 6 months

3 full tests with AI feedback, outside the weekly quota. No subscription needed; it stacks with one. Task sets and sections as on the free plan.

Buy a pack

At most 1 full test per day, on every plan; the day runs from midnight Toronto time (Eastern Time, daylight saving included).

Learn the format

A guide to every task, and plain answers.

Why are there no sample answers?

Memorised answers are easy to spot and hold your level down; the test provider warns that responses that are not your own can cancel a score. Feedback here starts from your own sentences and shows them revised.

Do the questions come from the test provider?

No. Every question is written for celpip9 on the published structure, and checked by independent gates before it is used. We are not affiliated with the test provider.

How hard are the practice tests?

Built to the published 2024 difficulty profile · Measured: not yet measured. Every test is built to the same profile, so none is labelled easier or harder. How difficulty is set

Is my level estimate my CELPIP® score?

No. It is an estimated level, based on the published Performance Standards dimensions, shown as a range.

Share your result

Help us publish the number nobody in this category publishes.

After your test, you can enter the level you received — only with your consent — so we can publish how close our estimate was.

  1. Take one full practice test free. It is Test 14 · Wrenfield, on everyday topics. No card.
  2. Sit your CELPIP® test. Keep practising here until then.
  3. Enter the level you received, if you choose. A separate consent on your Progress page. We publish totals across test takers only, never your result.
Start your free test