How many people will watch GTA 6 on launch day?

17 min readUpdated 13 November 2026

gta-6launchtwitchkickpredictiondata

About 2.4 million people at once, at the peak — across Kick, Twitch and YouTube combined.

That is a forecast, not a measurement, and the rest of this page is the working: what it is built on, which part of it is a guess, and the exact rule we will use to grade ourselves in November. We are publishing it now, fifty-five days out, so that nobody can accuse us of writing the prediction after the fact.

The second half of the forecast is the part we think is wrong everywhere else. Twitch, on its own, probably does not break the record — and that is not a claim about Grand Theft Auto VI being small. It is a claim about where GTA’s audience has moved.

Update, 13 November 2026 — six days out, the Twitch column has been re-measured, and it was too generous to Twitch. Nothing in the prediction table below has been changed: it stands exactly as published on 25 September, which is the whole point of having published it early. But the 42.7% Twitch share that the table’s third column is built on came from a single day of sampling in September, and a single day turns out not to be a safe thing to build a column on. Measured again over the twenty-four hours to 13 November 04:05 UTC, Twitch’s share of Grand Theft Auto viewer-hours is 33.1%, and within that one day it ranged from 23.6% to 51.9% depending on the hour. The detail is in a new section below, and it cuts in the direction of our own call rather than against it.

The prediction, in a form you can check

Prediction How it gets graded
Peak concurrent viewers, Kick + Twitch + YouTube 2.4M (range 1.5M–4.0M) Highest one-minute reading in live/peaks.json between 18 Nov 11:00 UTC and 23 Nov 00:00 UTC
Peak concurrent viewers, Twitch alone ≈1.02M Same window, Twitch series only
Does Twitch beat the 1,280,000 single-player record? No Twitch peak above or below 1,280,000
When the peak lands Evening of 19 November, 19:00–22:00 UTC Timestamp on the peak reading
Biggest single hour is not the American unlock hour (05:00 UTC, 19 Nov) Correct Compare the 05:00 hour against the day’s maximum

We will publish the measured figures against this table in the week after launch, including — especially — if they make us look silly.

The baseline this is built on

We sample every live Grand Theft Auto stream on Kick, Twitch and YouTube once a minute. Here is what the category does when nothing in particular is happening, over the twenty-nine complete days from 27 August to 24 September 2026 — 696 hourly points.

Measure Value
Mean concurrent viewers 283,370
Median hour 279,524
Busiest hour of the day (20:00 UTC), median 508,449
Quietest hour of the day (08:00 UTC), median 94,664
Highest single minute ever recorded 992,344 — 27 Aug 2026, 19:21 UTC
Total viewer-hours, 29 days 197,225,299

That 992,344 is the ceiling Grand Theft Auto has actually touched, and it took the Extended Look premiere to get there. It is also the number this forecast is scaled from, so it is worth being precise about it: it is the highest single-minute reading in our peaks file. Our post on the premiere quoted 959,078 at 19:25, which is the highest five-minute sample. Both are ours, both are correct, and the four-minute gap between them is worth 33,266 viewers — which tells you something about how sharp that spike was.

Why Twitch probably misses the record

The standing record for a single-player game on Twitch is 1,280,000 concurrent viewers, set by Hogwarts Legacy on 7 February 2023 during its early-access window — certified by Guinness World Records. GTA VI ships without an online mode, so that is the record it is chasing.

Here is the part only we can add. Over the last complete day of sampling — 04:00 UTC on 24 September to 04:00 UTC on 25 September, 6.78 million viewer-hours in total — GTA’s audience divided like this:

Platform Viewer-hours, 24h Share
Kick 3,605,495 53.2%
Twitch 2,897,234 42.7%
YouTube 277,862 4.1%

Twitch carries under 43% of people watching GTA. In February 2023, when Hogwarts Legacy set that record, Kick had been public for a matter of weeks.

So a 2.4 million peak, split the way GTA splits today, puts about 1.02 million on Twitch — roughly 260,000 short. For Twitch to take the record, it needs 53% of GTA VI’s live audience, ten points more than it holds in GTA right now. That is possible: GTA VI at launch is a single-player game with no roleplay, and Kick’s GTA dominance is built substantially on roleplay and on Arabic- and Portuguese-language creators. If the new category skews back toward Twitch, the record falls. We do not think it skews ten points.

This is why “will GTA 6 break the Twitch record” and “will GTA 6 be the biggest game launch ever streamed” now have different answers. They did not used to.

The peak will not be at the unlock

GTA VI does not come out at a single moment. It unlocks at local midnight, region by region, across nineteen hours — New Zealand first at 11:00 UTC on 18 November, Mexico last at 06:00 UTC on 19 November, with United States accounts unlocking at 05:00 UTC on the 19th. That is reported from store listings, not announced by Rockstar.

Lay that against the audience curve and something awkward falls out: the American unlock lands in the deadest part of GTA’s day.

GTA 6 unlocks in America during GTA’s quietest hours Median live GTA viewers across Kick, Twitch and YouTube by hour of day · 29 days to 24 September 2026 0 100k 200k 300k 400k 500k 181,108 US accounts unlock here 508,449 the day’s busiest hour 94,664 00 03 06 09 12 15 18 21 hour of day, UTC luciastream.com
Median concurrent viewers of live Grand Theft Auto streams on Kick, Twitch and YouTube by hour of day, over the 29 complete days from 27 August to 24 September 2026 (696 hourly samples). Unlock times are reported from regional store listings, not announced by Rockstar.

Five in the morning UTC is a 181,108-viewer hour, about a third of the day’s busiest. Nine hours of American streamers going live at midnight Eastern will lift it enormously in relative terms and still not produce the day’s biggest number, because the people who watch are asleep in the time zones that supply most of the audience. The largest single reading should arrive about fifteen hours later, on the evening of 19 November, when Europe, Brazil and the American evening overlap — the 19:00–22:00 UTC block that carries the daily maximum every day of the year.

If you are planning to watch the moment GTA streaming peaks, that is when. If you are a streamer planning a launch stream, the same table is a scheduling argument.

The one number we are guessing

Everything above is measured. This is not: we scaled the forecast at 2.4× our own all-time high.

The reasoning, so you can disagree with it precisely. A launch is bigger than a trailer for streaming, because thousands of channels sit inside the category for days rather than hundreds for twenty-six minutes. One twenty-six-minute presentation pulled the whole category to 992,344; a launch has no such ceiling on how many channels can be inside it at once. Pulling the other way, the unlock is smeared over nineteen hours, so there is no synchronised global moment of the kind that produced our 992,344 in one minute. One factor multiplies, the other flattens.

If the multiplier is Cross-platform peak Twitch at 42.7% Beats 1,280,000?
1.5× 1.49M 636k No
2.0× 1.98M 847k No
2.4× 2.38M 1.02M No
3.0× 2.98M 1.27M Line ball
4.0× 3.97M 1.70M Yes

You need GTA VI to draw four times the biggest live audience Grand Theft Auto has ever had before Twitch takes the record on today’s platform split. We would not bet against four times as a stretch case, which is why the range runs there. We would bet against it as the central estimate.

Re-measured six days out: the split is not a constant

The forecast above has one measured input doing a lot of work: the 42.7% of Grand Theft Auto viewing that sat on Twitch over the last complete day of sampling before publication. Every figure in the Twitch column is that share multiplied by a total. It was a single day’s measurement, presented as a property of the category, and six days from launch it is worth asking whether it holds.

It does not. Here is the same measurement, same method — viewer-hours integrated across every five-minute sample — over the twenty-four hours to 13 November 2026 04:05 UTC, beside the September figure the prediction was built on.

Platform Viewer-hours, 24 Sep Share Viewer-hours, 12–13 Nov Share
Kick 3,605,495 53.2% 4,061,329 61.0%
Twitch 2,897,234 42.7% 2,204,354 33.1%
YouTube 277,862 4.1% 389,752 5.9%
Total 6,780,591 6,655,435

Twitch is down 9.6 points on the same instrument seven weeks later, on a total that barely moved. The tempting thing to do with that is swap 33.1% into the sensitivity table and call the Twitch column fixed. That would be making the original mistake twice, because the interesting finding is not the new number — it is that there is no single number here to get right.

Within that one day, Twitch’s share of live GTA viewers ran from 23.6% to 51.9% across the twenty-three complete hours, a 28.3-point swing, on a predictable daily rhythm. Twitch leads the European afternoon and evening; Kick takes the late night and the small hours.

Why it moves that way we can only infer, and the inference is worth stating as one. The three largest Grand Theft Auto creators by average concurrent viewers are all Arabic-language Kick channels, and Arabic accounts for 15.2% of everyone watching live GTA across five channels — against English’s 34.4% across 1,182. The two biggest Twitch creators in the category are Spanish-language. A split that swings twenty-eight points between the European evening and 03:00 UTC is what you would expect from that cast list, but we have not measured each channel’s hours, so read the mechanism as a hypothesis that fits rather than a finding.

Hour, UTC Live viewers Kick Twitch YouTube
08:00 76,611 49.5% 23.6% 26.9%
12:00 146,904 54.9% 37.4% 7.7%
18:00 298,850 51.6% 41.3% 7.1%
19:00 354,625 42.0% 51.9% 6.2%
20:00 400,086 45.9% 50.4% 3.8%
21:00 591,181 63.9% 34.5% 1.7%
22:00 675,240 68.6% 30.4% 0.9%
00:00 563,909 70.8% 25.9% 3.3%
03:00 210,821 69.6% 23.7% 6.7%

Now put that beside what this page predicts. We said the launch peak arrives in the 19:00–22:00 UTC block on 19 November. That block is precisely where the platform split does its most violent moving — Twitch holds a bare majority at 19:00 and barely 30% three hours later. So “the peak lands at 19:00–22:00” and “Twitch holds 42.7% of it” were never independent claims, and the page treated them as if they were.

The part that matters for the record is the hour the maximum actually lands in, and there the split is unexpectedly steady. These are the six largest five-minute readings of the measured day, all of them inside the 21:50–22:15 window:

Time, UTC Live viewers Twitch Twitch share
21:50 694,864 210,670 30.3%
21:55 693,792 212,634 30.6%
22:00 692,599 212,363 30.7%
22:05 691,188 212,236 30.7%
22:10 688,828 211,761 30.7%
22:15 687,669 210,550 30.6%

Thirty per cent, within half a point, across the whole of the day’s maximum. The daily average swings wildly; the peak does not. And at the peak Twitch holds twelve points less than the share this page assigned it.

Which means the forecast’s central call gets safer rather than shakier. To take the 1,280,000 record at our 2.4× central case, Twitch needs 53.7% of a 2.38 million peak. On the September split it needed ten points more than it held. On the share it actually holds when GTA peaks, it needs twenty-three.

If the multiplier is Cross-platform peak Twitch at 30.3% (peak-hour) Beats 1,280,000?
1.5× 1.49M 451k No
2.0× 1.98M 601k No
2.4× 2.38M 722k No
3.0× 2.98M 902k No
4.0× 3.97M 1.20M No

On the peak-hour split, even four times the biggest live audience Grand Theft Auto has ever had leaves Twitch short. We are not moving the published prediction to match — it stays as written, and we will grade the 42.7%-derived 1.02M figure as the number we actually committed to. But if we are wrong on launch night, this table says it will be because the multiplier was too small, not because Twitch was underrated.

One thing is moving, and it is not the average

The other reason to re-measure now is that launch week has started to show up in the data — though not where you would expect.

Day Daily mean Highest hour
8 Nov 236,226 392,095
9 Nov 215,397 464,478
10 Nov 266,474 432,184
11 Nov 265,617 515,621
12 Nov 271,298 675,240

The daily averages are ordinary. 12 November’s mean of 271,298 sits above the 29-day mean of 248,005 and below 28 October’s 291,232 — nothing a countdown explains. The peaks are another matter: 12 November’s 22:00 hour read 675,240, the highest single hour in all twenty-nine complete days we hold, 16% clear of the previous best (581,566, 29 October). At minute resolution the same evening touched 703,930 at 21:56 UTC, the largest reading of the past seven days and still well short of the 992,344 all-time high from August.

So the audience is not getting bigger on average as launch approaches. Its evenings are getting taller. That is consistent with how the one event we have measured properly behaved — the Extended Look moved a single hour enormously and the day barely at all — and it is the shape a launch night would have to exaggerate for the 2.4 million to land.

What this section does not show

It is one day of hourly splits. The live endpoint publishes per-platform history for twenty-four hours only, so the hourly table cannot be averaged over weeks the way the original baseline was. The rhythm it shows is plausible and matches the September platform mix, but one day is one day, and a reader should treat the 23.6–51.9% range as a demonstration that the split moves rather than a measurement of how far it moves on a typical day.

We cannot say which channels drove the peak. Our board of live streams is a snapshot of now, not a history, so for 12 November’s 21:50 maximum we can see that Kick held 68.4% of it and not who on Kick held that. GTA’s upper tail is concentrated enough for that to matter: while this section was being written a single Kick channel held 12.1% of all live GTA viewers and 16.8% of Kick’s own total, and the entire Arabic-language audience — 15.2% of the category — sat on five channels, against English’s 34.4% spread across 1,182. So “Kick’s late evening” may be a handful of rooms rather than a platform-wide pattern. That distinction matters for the forecast and we cannot settle it with this instrument.

Viewer-hours and concurrent viewers answer different questions. The platform table integrates over a day; the peak table is a sum of per-stream counts at one instant. They are not interchangeable, and the 33.1% and the 30.3% are not two readings of the same quantity.

What would make us wrong

Hardest first.

The multiplier is a judgement. It is one person’s reading of two opposing forces, not a model fitted to anything. If it is wrong, every number on this page is wrong by the same factor, which is why the sensitivity table is above rather than buried.

We might undercount the first hours. Our collector finds GTA streams by category and by title search, and it polls for the names a platform might give the new category — “Grand Theft Auto VI”, “Grand Theft Auto 6”, “GTA VI”, “GTA 6” — every day, so a new category should be picked up the day it appears. Should is not the same as does. If a platform names it something we are not watching for, our launch-night figure is low until we notice.

We measure category-tagged streams, and that is not the same as “everyone watching”. Streams Charts put the Extended Look at 3.97 million concurrent across platforms; we recorded 992,344 for the same event, because most reaction co-streams sat in their channel’s usual category rather than in a GTA one. On launch night streams should be in the GTA VI category, so the two measures ought to converge. That is an expectation, not a measurement, and if it is wrong our number is the conservative one.

Concurrent viewers is not a count of people. It is a sum of per-stream viewer counts at one instant, so one person in three tabs is three, and it excludes anyone watching a platform we do not track. The methodology page is explicit about this, and so is the number itself.

The GTA V audience is in the total. The 2.4M is all live Grand Theft Auto viewing, GTA VI plus everything else, because that is what our instrument measures and what the 992,344 it is scaled from measured. Roleplay does not evaporate on 19 November — we have argued at length that it survives, because GTA VI has no PC version and therefore no FiveM.

Check us, and cite us

The live total is on the live board, updated every minute, and the running peaks are on the peaks page. Both read the same public JSON this post was written from.

If you are writing about launch night and want to quote a figure, the press page has the definitions and the endpoints, and this page is the prediction on the record:

luciastream.com predicted a peak of 2.4 million concurrent live Grand Theft Auto viewers across Kick, Twitch and YouTube during GTA VI launch week, published 25 September 2026 at https://luciastream.com/blog/how-many-will-watch-gta-6-launch-day/. Baseline: 696 hourly samples, 27 August – 24 September 2026. Platform split re-measured 13 November 2026 over the preceding 24 hours (6,655,435 viewer-hours): Kick 61.0%, Twitch 33.1%, YouTube 5.9%; Twitch held 30.3% at that day’s peak of 694,864 concurrent viewers. Methodology: https://luciastream.com/methodology/.

We will grade it in public.


Numbers in this post come from our own sampling — seehow we measure.