Nothing is decoded, so nothing is lost
Most online trimmers decode your video to pictures, throw away the frames outside your range, and encode the rest again. That is why they take minutes, why the result looks slightly softer than the original, and why trimming the same clip twice makes it worse both times.
This one never touches a pixel. A video file is a container holding already-compressed packets, and cutting it to a range is a matter of copying the packets you want into a new container. No decoder runs, no encoder runs, and the picture that comes out is the picture that went in, byte for byte. A ten-second cut from a 1080p file typically finishes in under a second, and the time it takes is bounded by how fast your disk reads rather than by how fast your processor encodes.
A useful side effect: because there is no encoder in the path, this works on machines that have no usable video encoder at all — including Firefox on Android, where every other video tool on this site has to refuse.
The keyframe problem, stated rather than hidden
There is one real cost to the fast path, and it is worth understanding because every lossless trimmer has it. Compressed video is not a sequence of independent pictures. Most frames describe only what changed since the frame before, so they are meaningless without the run of frames that precedes them. Only a keyframe stands alone.
That means a copy can only begin at a keyframe. Ask for a cut at 10.0 seconds on a file with keyframes every two seconds and the copy has to start at 8.0. Most tools round silently and hand you a clip that begins two seconds early, which reads as a broken trimmer rather than as a property of the format.
So this tool asks the file. As you move the start slider it looks up the last keyframe at or before your point and prints the exact figure — sometimes zero, often a fraction of a second, occasionally the full two. Nudging the start forward a little will usually find a closer one. If you need the cut frame-exact, you need a re-encode, and you should reach for a proper editor rather than a browser tab.
Four containers, and why not more
MP4, MOV, WebM and MKV are read; a QuickTime file comes out as MP4, because both are the same underlying format and MP4 is the one that plays everywhere. AVI, WMV, FLV, MPEG program streams and MXF are refused, and the refusal is deliberate rather than a shortcoming waiting to be fixed.
The library that would read everything is FFmpeg compiled to WebAssembly. Its core is GPL-licensed — its build enables x264 and x265, both GPL, and sending that binary to your browser counts as distribution, which would oblige us to publish this entire site under the same terms. It is also 30.7 MiB, which is larger than the maximum single file our host will serve. Neither problem has a workaround, so rather than promise a converter that fails on half of what it is given, the tool takes the formats it can genuinely handle and says so on the page.
A .mov holding ProRes or DNxHD is refused by name for a different reason: no browser has ever shipped a decoder for either, so there is no client-side route at all. Export H.264 from your editor first. More in video & audio tools.