Five untimed CELPIP Speaking Task 5 practice questions, each one showing both options as photographs and each with a spoken model answer. Your choice is already made; the job is to persuade a named person that it beats the one they want. Record yourself, reveal the model, then read every move it makes and why.
Task 5 is the longest and strangest thing on the speaking test, and the only one that runs across three separate screens. It is also the one candidates most often misunderstand before they open their mouths, because the shape of it is genuinely counterintuitive.
On the real test, two options appear: two properties, two venues, two pieces of equipment. You get 60 seconds to compare them and pick one, and nothing is recorded during that minute. Then the screen changes. You get 60 seconds of preparation, then 60 seconds to speak, and your job is to persuade somebody that your choice is the right one. Roughly three minutes end to end, and only the last sixty seconds are your answer.
Here is the twist that decides the task. On that speaking screen, the option you rejected is gone. It is not greyed out, not mentioned, not referred to again. In its place is a third option you have never seen, the one the other person wants. So the comparison you actually owe the rater is not the comparison you just spent a minute making. It is your choice against their choice, and the two options you were originally shown are now half irrelevant.
Two things about the practice questions below, so that nothing about them surprises you. The first is that each one shows both options as photographs with their details listed underneath, the way the real screen does, and hands you the rival's option in the same breath rather than making you sit through the choosing minute to meet it.
The second is that the choice is already made for you, and that is deliberate rather than a shortcut. Every question here opens the same way: you have chosen this one, and they want that one instead. Choosing is not what Task 5 scores. Nothing you do in that first silent minute is recorded, both options are winnable, and every mark on this task is earned in the last sixty seconds. Fixing the choice is also what lets a single model answer argue the option actually in front of you, instead of one model per branch for a laptop or a tent you never picked. What you lose is the ten seconds of deciding. What you keep is the whole scored part.
The live comparison is B against C. An answer that compares A with B is arguing about a card that is no longer on the screen, and it is one of the two ways this task is routinely failed.
The other way is subtler and more common: praising your own option for the full 60 seconds and never mentioning theirs. It feels like persuading. It is not comparing, and the task asks for both. A rater hearing a warm description of your laptop with no reference to the tablet has heard half a task.
Two more corrections worth having in advance, both of which contradict popular prep advice. The first: the rival is not always a family member. Official items use a co-worker, a friend, another supervisor, a father. Tone is scored, so the register you owe a supervisor you share a contract with is not the one you owe your own partner. The second: Task 5 is not always about money. Some official items contain no dollar figures at all, and blogs that tell you to always argue price first leave you stranded when the options differ on space, effort or dignity instead.
The 2026 CELPIP speaking format for the comparing and persuading task specifically.
Persuasion has a shape, and 60 seconds is enough for all of it if you know the moves.
Worth dwelling on the tone point, because it is worth real marks and costs nothing to get right. Task Fulfillment lists relevance, completeness, tone and length, and Task 5 hands you the tone in the prompt. The family paying for your computer get warmth, and can be told outright that their pen is lovely. Your own partner gets the register of two people deciding something together, not a case presented to a committee. A supervisor you will be running a Quebec contract with in the spring gets collaboration, because winning the argument and losing the colleague is not a win. Argue with all three the same way and one of them will cost you.
And notice what a good answer does with the arithmetic. It does not stop at ours is cheaper. It re-does the rival's numbers out loud: twenty-six hundred dollars buys nine certificates on paper, and a team on evening shifts will collect maybe four of them. One number is a fact. A number reframed is an argument, and only the second one persuades.
Five rivals and five registers: the family paying for your computer, your own partner, a friend, a co-worker and a supervisor you share a contract with. Two of the five are workplace decisions between equals, and one of them has no dollar figure in it anywhere. Every question shows you both options as photographs with the choice already made, so your minute goes on the argument rather than the deciding. Open one, record it with no clock running, listen back, then reveal the spoken model answer and every move it makes. Your recording stays in your browser: it is never uploaded and nothing here is scored.
A refurbished laptop against a brand new tablet, and the family helping pay for it want the tablet. The model grants that the pen is genuinely lovely before it ever mentions the software that will not install, which is what keeps the answer from sounding like stubbornness.
Practise this question
A dome tent against a hard shell rooftop tent, and your partner is sold on the three minute setup. Concede the three minutes in two words, then move the whole argument onto the walk-in sites a tent bolted to a roof can never reach.
Practise this question
Six days in Montreal and Quebec City against a guided coach tour of the Rockies, argued with a friend who has never seen the mountains. There is not one dollar figure in this question, so the comparison has to be built out of time, weather and freedom instead.
Practise this question
A used full size fridge against a new under counter one, and your co-worker is right about the noise. Six lunches for an office of fourteen is the number that answers them, but only once you have admitted the humming yourself.
Practise this question
A tutor at the workplace against the college evening course, and the other supervisor holds both the lower price and the only certificate. Conceding both of those in the first breath is the only honest way to open against a proposal that strong, and the close then redefines the goal from being certified to being able to answer the phone, all in the register you owe an equal you have to run the contract with afterwards.
Practise this questionNeither option is better, so do not deliberate. Pick the one you have more sentences about and give the remaining fifty seconds to the argument. Candidates who agonise over the choice arrive at the speaking screen with nothing prepared and a decision they still do not like.
Knowing it is coming is half the battle. Do not build your whole case on screen one against an option that is about to disappear. Build it on why your choice is good in general, then aim it at whatever appears.
Your family, your partner, a friend, a co-worker, another supervisor. It is written in the prompt and it sets your register before your first word. Arguing with a supervisor you share a contract with as though they were your partner is a Task Fulfillment error, not a personality.
Your partner says three minutes, and they are right about three minutes. So talk about the walk-in site the car cannot reach and the Saturday the car cannot move, not about setup times. Meeting an argument where it is strongest is polite and slow. Meeting it where it is silent wins in sixty seconds.
"Seven in the morning, on the coach. Forty minutes at the lake, then back on the coach." Playing their option forward hour by hour is more devastating than any adjective, and it is concrete language doing persuasive work.
"I would rather hunt for a plug than hunt for a computer that runs my course." A parallel that puts the two futures side by side lands better than any summary, and it uses the last five seconds instead of wasting them.
CELPIP Speaking Task 5 is Comparing and Persuading. You are shown two options, you choose one, and then you persuade somebody else that your choice is better than the one they want. It runs as three screens and takes about three minutes in total. It is the fifth of the eight speaking tasks in the 2026 CELPIP format.
Because the choosing and the speaking are separated. Screen one gives you 60 seconds to compare the two options, with no recording at all. Screen two gives you 60 seconds of preparation, and screen three gives you 60 seconds to speak. Roughly three minutes in total, and only the last 60 seconds are recorded.
It vanishes, and this is the detail that catches most candidates. On the speaking screen only two options are shown: your choice, and a third option that somebody else wants, which you have never seen before. The option you rejected is gone and is never mentioned again. So the comparison you owe the rater is between your choice and their choice.
It varies, and the variation matters. Official items use a family member, a co-worker, a friend, another supervisor and a father. Tone is scored under Task Fulfillment, so the register you owe a family member is different from the one you owe another supervisor. Read who they are before you decide how to argue.
No. Popular advice says always argue price first, and the official item bank contradicts it: some items contain no dollar figures at all. A fully qualitative Task 5 is common, and candidates who arrive hunting for a price argument are stranded when there is not one. Build the comparison out of whatever the options actually differ on.
The computer picks one for you and the test moves on. You are not penalised for the choosing part itself, since nothing is recorded there, but you then have to persuade somebody in favour of an option you did not select. Choosing quickly and deliberately is worth more than choosing well: neither option is better, only easier to argue.
Yes, and it is one of the strongest moves available. Naming the tired battery, or the fact that your tutor has no certificate to hand out, and then answering it is what makes the rest of your argument credible. An answer that pretends its option has no drawback sounds like an advertisement, and it gives the listener nothing to accept.
Because choosing is not the skill Task 5 scores. Every question here opens with the choice fixed: you have chosen one option, and the other person wants the other. On the real test you spend 60 silent seconds picking, but nothing in that minute is recorded or marked, and the persuading is where every mark lives. Fixing the choice also lets one model answer argue the option in front of you, instead of needing a different model behind every branch you might have taken.
No. There is no timer and no evaluation anywhere in the practice course. You record yourself so that you can hear yourself, the recording stays in your browser and is never uploaded, and then you reveal a spoken model answer with its structure, key phrases and common mistakes.
Yes. All 5 questions, both photographs on every one of them, all 5 spoken model answers and every explanation on this page are free, with no account and no credit card required.
Yes. Everything here follows the 2026 CELPIP speaking format: eight tasks, a timer that runs per task and resets on every advance, and Task 5 running as three screens with 60 seconds to choose, 60 seconds to prepare and 60 seconds of recorded speech.
Task 6 also gives you a minute to prepare and a person to talk to, except now the situation is awkward and you speak to them rather than about them.
Two people you could speak to, and you must pick exactly one. The whole task is scored on tone.
Learn Task 6 → 7Argument again, but with nobody to persuade and nothing on screen to lean on.
Learn Task 7 →Every task, every question, every model answer, on one page with the topics listed in full.
See the samples →All 8 tasks back to back, with the three screen flow running on the real clocks.
See the speaking section →Five arguments, no clock, no score. Record your case, then check whether you ever actually mentioned their option. That check is the task.
