Video models

Veo 3.1 or Sora 2?

Both of these can talk. Is this one beautiful shot, or a scene with a joke in it?

The short answer

Google

Veo 3.1

Veo 3.1 is the cinematic one: cranes and push-ins that read like a real operator, faces that hold through the move, and sound recorded in the same take.

Sound in the same take
Voices, room tone and effects come back with the picture, in sync.
Camera that means it
Dollies, cranes and slow push-ins that read like a real operator.
Faces that hold
Close-ups that keep skin, eyes and expression steady through the move.

The tailor's verdict

Prompt to tryAn old tailor pins a navy jacket on a young man, steps back and says 'now you look like yourself'

OpenAI

Sora 2

Sora 2 is the one that lands a beat — timing, reaction and a punchline, inside a world where objects keep their weight from the first second to the last.

Scenes with a punchline
Timing, reaction and a beat at the end — the stuff that gets shared.
Worlds that make sense
Objects keep weight and cause keeps effect across the whole clip.
Talking to camera
Characters who speak their lines and mean them.

The honest review

Prompt to tryHolding her phone in the bathroom, a woman tells you honestly why she switched moisturisers

The next question

Either way the price is quoted before the render runs, and a render that fails is never charged for. If you would rather not choose, the studio picks for you on every ordinary render — this page is for when you would.