Distribution is not delivery

YouTube 4K bitrate and client delivery.
The quality question is only one of four.

A client screener on YouTube fails on fitness for purpose before it fails on bitrate. YouTube states that it always re-encodes, that a new upload is processed in low quality first, and that it converts your file for playback across devices and networks. Those three sentences, all published by YouTube, decide the question. Access control decides the rest.

See the review room

Video hosting for filmmakers · 5 GB free · Paid plans from USD 9/month

By uncompressed.io · Updated September 2026

Frame the decision properly

This is not only a quality question,
it is a fitness-for-purpose question.

The search that brings people here is about bitrate, and bitrate is a real part of the answer. It is one part of four. A client screener has a job to do: present a specific cut, at the quality you signed off, to a named person, on a page you control, and collect a decision. Ask whether YouTube is built to do that job, rather than whether its picture is good enough, and the answer stops being a matter of taste.

YouTube is a distribution platform. Everything about it is designed to get a video in front of as many viewers as possible, on as many devices and connections as possible, at as low a cost per view as possible. Those are the right goals for distribution and they are the wrong goals for a screener, where there is one viewer, one device, one cut, and the only cost that matters is the cost of a client seeing the work worse than you graded it.

The four reasons below are built on sentences YouTube publishes itself. We have deliberately not built the argument on a delivered-bitrate number, because YouTube does not publish one at any resolution, and a page that invents one is not worth reading. The number question is answered separately on how much does YouTube compress video, where the measurement is labelled as a measurement.

Always

re-encoded

YouTube's own word for what happens to every upload

Low first

processing order

high resolutions arrive hours later

Not you

who picks the rendition

the device and the network do

Zero

published delivery bitrates

at 4K or at any other resolution

Reason one

Every upload is re-encoded,
and the first hours are the low-quality pass.

YouTube states it plainly: “YouTube always re-encodes videos to optimize their playback quality.” There is no setting, no tier and no upload format that avoids it. Your export is a source file for YouTube’s encoder, not the file anyone watches.

The second published sentence is the one that costs people work. On its processing page, YouTube writes that “when you upload a video, it will initially be processed in low quality”, that higher qualities are processed afterwards, and that a video “may seem to be missing higher qualities for several hours”. Its own worked example is a 60-minute 4K 30 fps video taking up to 4 hours to finish high-resolution processing, on the reasoning that 4K videos are four times larger than 1080p videos.

Now put that on a delivery day. You finish the grade at four in the afternoon, upload, wait for the progress bar to clear, and send the link at five. The client opens it on a phone in a taxi. What they watch is whatever pass has finished, and their first impression of six weeks of work is formed on it. Nothing about that is a YouTube failure: it is a distribution platform doing exactly what it says it does, on a job it was never asked to do. It has cost people jobs, and it is entirely avoidable.

If you do use YouTube for a screener, at minimum

Upload the night before, not an hour before. Check the quality menu on the watch page yourself before you send the link, from a different network than the one you uploaded on. Follow YouTube’s published upload recommendations rather than guessing: MP4 with the moov atom at the front of the file, H.264 High Profile, closed GOP, CABAC, and 35 to 45 Mbps for 4K SDR at 24 to 30 fps. A cleaner source will not stop the re-encode, but it gives the encoder less to throw away.

Reason two

You do not choose what the client sees;
the client’s connection does.

YouTube states that your video “will be converted to the highest resolution available to ensure successful playback on different devices and networks”. Read that as an engineering goal rather than a marketing line and it is exactly right: successful playback everywhere is the product. It also means the rendition your client receives is selected by their hardware and their bandwidth, not by you.

For an audience, that is a feature. A viewer on a train should get a watchable stream instead of a spinner. For a screener it inverts the relationship: you spent the week deciding exactly what the film looks like, and the last decision in the chain is made by a hotel wifi network on the client’s behalf. You will not know which rendition they got. Neither will they, unless they open the quality menu, and clients do not open the quality menu.

It is worth being precise about what is public and what is not. YouTube publishes recommended upload bitrates and, separately, live ingest bitrates for streaming to YouTube. Both describe what you send. Neither describes what a viewer receives. The only delivered figures YouTube publishes anywhere are audio at 256 kbps for Premium music videos and Art Tracks, and a non-numeric note that 1080p Premium is an enhanced bitrate version of 1080p. For the delivered video bitrate at 4K there is nothing, and that absence is itself the point: a screener workflow should not rest on a number the vendor declines to state.

YouTube

Vimeo

Recommended4K upload

35 to 45 Mbps SDRat 24 to 30 fps

30 to 60 MbpsCRF 18 or below

No recommendation.Upload the exportyou want played

Published 4Kdelivery bitrate

None published,at any resolution

Up to 22 Mbps H.264,up to 16 Mbps HDR

The file you uploaded

Re-encodesyour file

Yes, always,in its own words

Yes, into severalformats

No transcode,no adaptive ladder

Who picks therendition

Device and network

The transcode ladder

There is only one file

Wait beforehigh quality

Up to 4 hours for a60-minute 4K 30 fps video,YouTube's own example

Not published

None: the bytes areserved as uploaded

The honestcaveat

Plays on everything,everywhere, free

Published ceilings aremaximums; output canbe lower

A slow connection buffersinstead of dropping quality.Playback needs the viewer'shardware decoder

YouTube figures from its recommended upload encoding settings and its video-processing page; Vimeo's recommended upload bitrates and its CRF guidance come from its video and audio compression guidelines; its delivery ceilings come from its playback-resolution guidelines, which state the listed bitrates are a maximum and output can be lower. All fetched 4 September 2026. For scale on what a delivery ladder is designed to carry, Apple's HLS authoring specification puts its top 4K HEVC rung at 16,800 kbps SDR: streaming as an industry is built around a ceiling in that region, an order of magnitude under a 150 Mbps master.

Reason three

A YouTube watch page
is a YouTube page.

This one is not about pixels at all. It is about the frame around the work, and it is the reason a lot of studios moved screeners off YouTube long before they cared about bitrate.

A watch page is a page on a consumer product built for discovery. It carries YouTube’s branding and player chrome, it sits inside YouTube’s recommendation surfaces, and it lives on a platform that carries advertising. None of that is a criticism: it is what a free distribution platform is, and it is why the platform is free. It is simply a different object from a page whose only job is to present one cut to one client.

The practical consequence is small and constant. Your screener arrives wrapped in someone else’s identity. The client’s attention has somewhere else to go the moment the film ends. And the artefact you handed a paying customer, the thing they will show their own boss on Monday, is a link to a consumer video site rather than a page with your name on it. Studios pay for review platforms partly for the feature list and partly for that.

Reason four

A screener link should open for one person, close on command,
and never turns up in a search.

Start from the promise you make to the client rather than from any platform’s settings page. Most screener obligations reduce to three properties, and they are worth writing down before an embargoed cut leaves the building.

Only the client can open it

Not everyone the client forwards it to. Not anyone who finds the URL in a shared inbox. A link that is secret is not the same object as a link that is gated.

You can revoke it

The review window closes, the deal changes, the talent asks for the cut to come down. Access should end when you decide it ends, not when you remember to delete something.

It never surfaces publicly

No search result, no recommendation, no channel page. An unreleased spot with a client's brand in it should be invisible to everyone who was not invited.

Here, the first two are a password on the video and links that expire on their own. That is the whole of it, described accurately: a per-video password, and short-lived links. There is no multi-video client portal, and this page is not going to invent one. If your workflow needs a branded hub with every client project in it, that is a real requirement and we do not meet it today.

In the vault, which every plan has, the model changes shape entirely. The master goes dark on the web and is reviewed only in the macOS app, encrypted at rest, with capture restrictions on supported macOS configurations, a per-session watermark on the frame and an invisible per-person forensic mark. Access is revocable in one click and every open is written to a tamper-evident log. That is for unreleased features and embargoed campaigns, not for a Tuesday rough cut, and it is worth knowing which of the two you are actually being asked for.

What to send instead

A page with your name on it,
beside the file they are being asked to approve.

The replacement for a YouTube link is not a different video site. It is a workflow: one page that streams the exact export, holds the deliverable, carries the notes, and knows which version is which.

Stream the exact file

Upload the export you graded, up to 5 TB per file with no codec allowlist, in H.264, HEVC or AV1 if a browser has to play it. The player streams those bytes: no transcode, no adaptive ladder, no second version of your film. A 150 Mbps export plays at 150 Mbps, subject to the viewer's hardware decoder, and there is no processing window to wait out before it does.

Put a password on it and an expiry on the link

The client opens it. Nobody else does. When the review is over, the link stops working. Neither of those is a setting the client has to understand.

Put the deliverable next to the stream

The stream is for judging. The file is for using. ProRes 422, ProRes 4444 and DNxHR are hosted for download rather than streamed, because browsers do not decode them, so the client watches the HEVC or H.264 cut and downloads the master from the same page.

Take the notes on the frame

Versions with per-version approvals, and timecoded comments, so a note lands on the frame it is about instead of arriving as 'around the middle, the blue thing' in an email at eleven at night. Approve or reject each version on the record.

Every behaviour in that list is shipped and described by what it does. Captions are generated by a speech-to-text model when you ask for them. Bandwidth is never metered on any plan, including the free one, so a client who watches a 4K cut eleven times does not create a bill or a cap.

The honest part

Keep the film on YouTube for the audience,
and it is not close.

Nothing on this page is an argument for leaving YouTube. It is an argument for using it for the job it wins, which is reaching people.

For distribution, YouTube is unmatched and it is free, permanently, at any scale, with an audience that is already there and a player that works on every device anyone owns. It accepts .MOV, ProRes, DNxHR, HEVC and CineForm among others, up to 256 GB or 12 hours per file. No hosting product replaces that, and any page that tells you otherwise is selling you something. The finished film, the case study, the reel, the behind-the-scenes cut: all of that belongs on YouTube, and putting it there is the single highest-return thing most production companies do with their work.

The screener is a different job with a different customer. One named viewer, a fixed cut, a decision to collect, an obligation to keep it off the open internet, and a deadline that does not accommodate a four-hour processing pass. Run both. Publish to YouTube for the audience, and keep the screener somewhere that answers to you. The two are not in competition, and pretending they are is how comparison pages become useless.

What it costs

YouTube is free,
so name the price plainly.

A page arguing against the free option owes the reader the bill. Bandwidth is never metered on any plan here, including the free one, so no per-view charges apply and the only number to size.

YouTube: free

Unbeatable for reach, forever, at any volume. Files up to 256 GB or 12 hours. Always re-encoded, low quality first, and the rendition is chosen for your viewer.

Starter: free

5 GB of hot storage and one film at a time, no credit card. Enough to run a real screener through a real review cycle before you decide anything.

Lite: $9 a month

Or $99 a year. 100 GB hot plus 100 GB cold. A working freelancer's screener library: a year of cuts at a high-bitrate export, with the delivered masters coming and going.

Pro: $39 a month

Or $429 a year. 250 GB hot plus 750 GB cold. A small company running several clients at once, with masters parked beside the streams.

The comparison is not hosting against hosting. It is a free distribution platform against a paid delivery workflow, and the money buys four specific things: no re-encode between your export and the client’s screen, no processing window before the film is at full quality, a page with your name on it instead of a consumer product’s, and access you can gate and revoke. If a job never needed those, YouTube is the correct answer and you should keep using it.

Questions

Frequently asked

What bitrate does YouTube deliver at 4K?

YouTube does not publish one, at 4K or at any resolution. It publishes recommended upload bitrates, 35 to 45 Mbps for 4K SDR at 24 to 30 fps, and it publishes live ingest bitrates, which are what you send to YouTube rather than what a viewer receives. Every delivered-bitrate figure in circulation is somebody's measurement, not a Google figure. Our own measurement and the reasoning behind it are on the companion page, how much does YouTube compress video.

Why does my 4K upload look soft for the first few hours?

Because YouTube says that is how processing works. In its own words, when you upload a video it will initially be processed in low quality, higher resolutions are processed afterwards, and the video may seem to be missing higher qualities for several hours. YouTube's own worked example is a 60-minute 4K 30 fps video taking up to 4 hours to finish high-resolution processing. Send the link an hour after upload and your client may be judging the low-quality pass.

Does uploading a higher-bitrate master fix it?

No. YouTube states that it always re-encodes videos to optimize their playback quality, and that your video is converted to the highest resolution available to ensure successful playback on different devices and networks. A bigger upload gives the encoder a cleaner source, which is worth doing, but it does not change the fact that the client watches a rendition YouTube chose, on the connection and device the client happens to have.

Is unlisted good enough for a client screener?

It depends on what you promised the client. The three properties a screener usually needs are that only the named recipient can open it, that you can revoke it, and that it does not surface publicly. Here, that is a password on the video and links that expire on their own. Judge any visibility setting against those three, in writing, before an embargoed cut goes anywhere.

So should I stop using YouTube?

No. Keep the film on YouTube for the audience. YouTube is the best distribution surface that exists, it costs nothing, and no hosting product replaces it for reach. The argument on this page is narrower: publishing to an audience and showing a cut to a paying client are two different jobs, and the platform that wins the first one is the wrong tool for the second.

What does it cost to keep screeners somewhere else?

Starter is free: 5 GB of hot storage and one film at a time, no credit card. Lite is $9 a month or $99 a year with 100 GB hot plus 100 GB cold. Pro is $39 a month or $429 a year with 250 GB hot plus 750 GB cold. Bandwidth is never metered on any plan, including the free one, so no per-view charges apply. YouTube is free, so the honest comparison is that you are paying for control, not for hosting.

What plays here and what does not?

Up to 5 TB per file with no codec allowlist, streamed as the exact bytes you uploaded, and a browser plays H.264, HEVC or AV1. ProRes 422, ProRes 4444 and DNxHR are hosted for client download rather than streamed, because browsers do not decode them. Camera RAW can be uploaded but never streams in a browser: the compatible macOS app plays R3D, BRAW and N-RAW natively, and browser review needs an H.264, HEVC or AV1 export. Playback of anything high-bitrate depends on the viewer's hardware decoder, and there is no adaptive ladder here, so a slow connection buffers rather than quietly dropping to a mushy rendition.

Send the screener from a page that is yours

An unbranded watch page streaming the exact file you exported, a password on it, the deliverable downloadable beside the stream, and versions with per-version approvals and timecoded comments. Start free with 5 GB.