Clip length: the unit a line has to fit inside
Clip length is how long one generation runs. Published figures here range from a menu of four, six or eight seconds to a range of two to thirty, and the arithmetic against speech is short enough to do while writing. As of 2026-09-12.
| Model | Published clip length | What it means for a line |
|---|---|---|
| MiniMax | Four to fifteen seconds | A range, chosen on the call |
| Sora 2 | Sixteen or twenty seconds, extendable to 120 | Extensions add joins, not continuity |
| Veo | Four, six or eight seconds | A menu that reaches the script |
| Wan 3.0 | Two to thirty seconds | A range wide enough to need cutting |
Inclusion rule. Entries whose documentation publishes a duration for one generation. A ceiling on supplied audio, or on an account allowance, is a different figure and is recorded elsewhere. Order. Alphabetical by model name.
1Three words a second, and the sum nobody does
At roughly three words a second a six-second shot holds about eighteen words including the pauses at either end. Dialogue written without that constraint runs longer, and the result is a rushed read or a line that does not fit.
Doing the arithmetic while writing is far cheaper than discovering it at generation, because the fix is a rewrite either way and at generation it also costs a call.
2A menu constrains a script and a range does not
Where the options are fixed, dialogue has to be written in those units, and everybody involved can see the shape of the thing being made. Where the range is wide, the decision is postponed and the rework arrives later.
Neither is better in the abstract. Which one suits a production depends on whether the script or the schedule is the tighter constraint, and the choice is easier to make deliberately than by accident.
3An extension is an assembly rather than a take
Reaching two minutes by adding twenty seconds at a time produces a sequence of joins. Each join is a place where a voice, a level or a room tone can move, so the published total describes what can be stitched together.
Read as a budget for one beat the figure is useful. Read as a promise of two continuous minutes it will disappoint, and the documentation does not say which reading it intends.
4Sources read for this entry
This page defines the term and logs what each vendor documents about it, each figure read on 2026-09-12. Related: On-screen speaker, Native audio.