Compress a video to 5 MB

Where a 5 MB limit still comes from

Five megabytes is a limit written before anyone expected video to be attached to anything. It survives in forum software configured a decade ago, in wikis, in small bug trackers, in the file field on a university or council web form, and in self-hosted tools whose settings nobody has revisited since the install.

A lot of it traces back to a default that got nudged rather than reasoned about. PHP ships with upload_max_filesize = 2M and post_max_size = 8M, and the usual response to somebody complaining about that is to raise it to a round number. Five is a common place to stop. If a form rejected a small video without naming a size, 5 MB is worth trying before you assume the wall is at 8 or 10.

Older mail gateways cluster here too. An internal relay configured in 2009 does not get audited often, and the person who set the attachment cap was thinking about spreadsheets.

What actually fits in 5 MB

Duration decides this one completely. Five megabytes is 41.9 million bits. Across fifteen seconds that is a real budget and the picture can be genuinely good. Across five minutes there is nothing left once the audio has taken its share.

The table is generated from the same arithmetic the compressor runs, with the audio allowance already subtracted. At this ceiling read the right-hand column as a hard limit rather than a recommendation. It is what the budget can carry on ordinary footage, and grain or fast motion will not reach it.

Megabytes, or mebibytes

Everywhere else on this site the difference between the two is pedantry. At 5 MB it is the difference between a file that uploads and one that does not.

This tool treats a megabyte as 1,048,576 bytes, which is what your operating system means when it reports a file size. Plenty of servers check the upload against 5,000,000 instead. The gap is roughly 243,000 bytes: under five percent, and completely invisible right up until it is the reason for the rejection.

The fix is not to work out which definition the server is using. Set the target to 4 MB. That costs you a fifth of your budget, which at this size genuinely hurts, and it removes the whole category of failure. A file that gets rejected is worth nothing no matter how good it looked.

The audio track is not a rounding error here

On every other page on this site the audio allowance is a small subtraction from a large number. At 5 MB it can be most of what you have. The compressor scales the allowance down as the total budget shrinks, but it does not scale it to nothing, and by the five-minute row the audio is close to half of everything available.

Setting audio to Strip audio hands that share back to the picture, and at five minutes it roughly doubles the video budget. It is worth knowing what doubling gets you: still under what H.264 needs to hold a coherent picture together at any frame size. Stripping the audio at this ceiling is a real lever and it is frequently not enough on its own.

If the audio is the point — someone speaking, a recording of a call, a clip where the joke is the sound — do not strip it. Cut the clip instead. A minute of audible speech at 480p is worth more than three minutes of a silent slideshow.

Trim first, compress second

Past roughly three minutes, 5 MB has no answer. Not at 360p, not with the audio removed, not at any setting in the tool. Below about 150 kbps the H.264 encoder cannot hold a picture together at any frame size, which is where the table stops offering a resolution: you get blocking, smearing, and frames that visibly rebuild themselves after every cut.

So the useful move at this ceiling is to shorten the clip, and the scale of the effect is worth being blunt about. Fifteen seconds of the same footage holds 1080p inside 5 MB. Forty seconds holds 720p. The distance between impossible and genuinely good is a cutting decision, not a settings decision.

The tool takes a start and an end point, so you can make that cut here: drag the trim handles to the part that matters and the bitrate is solved from the trimmed length rather than the whole recording. If you need to remove something from the middle instead, cut it in whatever produced the file — Photos on macOS and iOS, Clipchamp on Windows 11, the gallery app on every phone — and bring the short version back. Either way, once the clip is short the ceiling stops being interesting.

Questions about the 5 MB target

Other size targets

• Compress to 8 MB The classic chat and forum attachment wall. Tight enough that duration decides everything.
• Compress to 10 MB The default cap on a huge number of upload forms — and 25% more room than 8 MB.
• Compress to 16 MB The WhatsApp video ceiling — and the one target where the platform re-encodes after you.
• Compress to 20 MB The default in helpdesks, CMS media libraries and upload widgets. Rarely announced.
• Compress to 25 MB The Gmail attachment limit — with the base64 overhead that makes 25 MB less than it looks.
• Compress to 50 MB Roomy enough that 1080p survives on short clips. The first target that stops hurting.
• Compress to 100 MB Generous enough for real 1080p. Also GitHub's hard per-file push limit.
• Compress to 200 MB Big enough that the real question is your viewer's patience, not the platform's limit.
• Compress to 500 MB Discord Nitro's per-file ceiling, and the point where the browser becomes the bottleneck.

Compressing for a specific platform instead? Discord, WhatsApp and email each have their own page.

Or start from the file you have. Whether 5 MB is a stretch or a formality depends on what is already inside the container: MP4 , MOV , MKV , WebM , AVI , WMV , FLV , M4V and MTS / M2TS .