How Long Does It Take to Make an AI Video? Our Times

An AI video from photos usually takes 3 to 20 minutes to generate, depending on the model, the quality and the length. In our own logs, a 12-second clip on Seedance 2.0 Mini at 720p was ready in about 3.7 minutes, and a 12-second Hotel Lobby video on Wan 3.0 at 480p in about 7.8 minutes. The same video at 720p or 1080p took closer to 20.

Updated October 19, 2026

Render times

How long an AI video takes

Our real numbers.

Fast model
480p first
Close the tab

What you fill in before the clock starts: the actual RapVideo editor. The times below come from our own renders.

These are medians from 126 finished paid renders on RapVideo between October 4 and October 18, 2026, timed from the moment someone pressed Create to the moment the MP4 could be downloaded. Your setup takes a few minutes more: choosing photos, writing a topic, picking a song.

Our render times, model by model

Model	Quality	Length	Typical	9 in 10 done within	Renders
Seedance 2.0 Mini	720p	5 s	2.9 min	14.3 min	8
Seedance 2.0 Mini	720p	12 s	3.7 min	6 min	17
Wan 3.0	480p	12 s	7.8 min	12.9 min	60
Wan 3.0	480p	15 s	10.6 min	12 min	15
Wan 3.0	720p	12 s	18.8 min	22.4 min	7
Wan 3.0	1080p	15 s	16.8 min	22.1 min	6

"Typical" is the median: half the renders were faster. The last column is the 90th percentile. Groups with fewer than six renders are left out, and the 720p and 1080p rows rest on few renders, so read them as a rough guide. The first 32 seconds or so (median) go to preparing the photos; the rest is the model working. The free 2-second preview took about 4.8 minutes across 96 renders.

What makes an AI video slow or fast

The model. Seedance 2.0 Mini is built for speed; Wan 3.0 follows the Hotel Lobby reference clip and keeps its track, and takes about twice as long.
Quality. Going from 480p to 720p more than doubled the time on the same model and length.
Length. More seconds means more frames. On Wan, 15 seconds took a few minutes longer than 12.
A reference clip or song. The model has to match its moves or lip-sync to it.
Queue. Providers run many jobs at once; at busy hours a render waits its turn. That is why the slowest renders in the table take two to five times the typical one.

Some newer models are much faster. This creator reports a model making five seconds of video in under three seconds; that is their claim, and we have not measured it. The other clip is the feeling most people know.

From TikTok, credited to the creators. Hover over a clip (or tap it) to play; tap a name to see the post.

How to get your video sooner

Pick Seedance 2.0 Mini when you want an AI rap, an effect or another stage; it was the fastest in our logs.
Make it at 480p or 720p first. Go to 1080p only for the version you will post.
Start with a shorter length to check the photos work, then make the full one.
Leave the tab open or come back later. The video is saved in My videos, and the tab title says when it is ready.

If a render fails, its credits come back on their own: 11 of 137 paid renders in the same two weeks failed and were refunded. You can run up to three at once.

How long a whole AI video project takes

Generating is only part of it. A single clip for TikTok is about 10 minutes from photos to download. A longer video made from several clips takes an hour or more, because each clip renders on its own and you still cut them together; see how to make AI videos longer. Writing the words takes time too, if you want your own: how to make a rap about yourself.

Make a video and time it

FAQ