The Auckland MMI is asynchronous now: the Kira Talent format, explained

The format most preparation pages describe no longer exists

Search for Auckland MMI preparation and you will be told, on several commercial pages, that you will walk a circuit of eight rooms, meet an interviewer in each, read a prompt on the door for two minutes and then talk for six. That is a real MMI format. It is not the one Auckland runs.

There is no interviewer in the room, because there is no room. You sit at your own computer, a video plays, and the platform records you. Nobody reacts, nobody prompts you, and nobody asks a follow up. That changes almost everything about how you would sensibly prepare, which is why it is worth getting the mechanics exactly right before you practise anything.

One station, second by second

UoA sets out the sequence in its July 2026 FAQ PDF. Every assessed station runs the same clock, and the whole station is three and a half minutes.

1. 1A short video plays of somebody reading the scenario aloud. You watch it, and the same text sits underneath the video and stays on screen. There is one question per station, and no follow-ups: your answer is recorded, so nobody is there to ask a second one.
2. 230 seconds of preparation. Your camera is on and you can see yourself, but this part is not recorded. You may make handwritten notes on one blank sheet of paper. That permission is explicit, and pre-prepared notes of any kind are not allowed.
3. 3Three minutes of recorded response. You start speaking when the response time begins. The scenario text stays on screen, you can still see yourself, and there is a timer at the top of the screen counting down.
4. 4Ten seconds of transition before the next station. You can skip it if you want to.

Seven scenario stations, taking about 30 minutes altogether, and then the ID and Safety Check steps at the very end: showing your primary and secondary ID and answering the safety questions. Those steps sit after the seven and are not one of them.

One take. No pause. No re-record.

UoA explains why in its FAQ: one opportunity to respond "helps us understand their natural reasoning and communication style". Whatever that does for the university, for you it means the first thirty seconds of your answer are the ones that ship. There is no version where you get to the end, dislike it, and go again.

This is the single hardest thing about the format and it is almost entirely a practice problem rather than a knowledge problem. Reading a model answer does not rehearse the part that goes wrong, which is the moment forty seconds in when you realise the structure you chose does not fit and you have two minutes twenty to talk your way out of it with nobody helping.

The scenario stays on screen, and there is a visible timer

Two details that quietly make the format easier than people assume. First, you are not answering from memory: the scenario text remains displayed while you record, so you can glance back at it. Second, the platform shows you the time you have left. UoA advises using it.

That is a real instruction and it has a real consequence: a strong ninety second answer is an incomplete answer. Half your practice should be about having enough to say, not about saying it more cleverly. We wrote a whole page about pacing a three minute answer.

Seven people mark you, and none of them is a machine

Worth sitting with, because it cuts against the instinct that a recorded format must be machine marked. It is not. Seven separate humans see one answer each, at random, and none of them sees your other six. There is no cumulative impression to build and no rapport to bank. Each station is a fresh stranger.

Which also means a bad station is genuinely contained. The person who marks station four never saw station three.

What you need in the room, and what you cannot have

• A desktop or laptop with a working webcam and microphone, or wired headphones. UoA states you cannot use a tablet or a mobile phone. If you have spent three weeks practising on your phone, you have been practising on a device you are not allowed to use.
• Paper and a pen. Handwritten notes during the 30 seconds are explicitly permitted.
• Two forms of ID, primary and secondary, for the station at the end.
• Nothing else. No books, no pre-prepared written or electronic notes, no cell phone, no smartwatch, no additional technology, and no recording devices.

The platform also offers accommodations available to everyone without prior approval: closed captioning on the recorded video, zoom support up to 200 percent, keyboard navigation, audio notifications of timed questions for screen readers, and the handwritten notes permission above.

So what does this format actually reward?

UoA is unusually direct about this, and it is the most useful paragraph in any of its four PDFs.

And, on the thirty seconds: reviewers "recognise that you have only had 30 seconds to think about the scenario so they are not expecting you to cover everything, they are looking for your ability to think on your feet". You are also explicitly allowed to change your mind mid answer, and UoA says being prepared to shift your position shows flexibility.

Put those together and the format is a delivery test wrapped around a reasoning test. Which is exactly the thing a bank of model answers cannot train, and exactly the thing repeated timed reps on camera can. What is actually being scored, domain by domain and attribute by attribute, is here.

The dates, briefly

The 2026 general MMI is on Tuesday 8 September 2026, at 11.00AM. Every applicant sits it at the same time. Invitations are sent on 26 August 2026, and scores are emailed around October 2026. The full timeline and what to have ready on the day is here.