AI video model

Sora 2

OpenAI's flagship model for physically grounded, controllable clips

Sora 2 is OpenAI's flagship video model, built for one clear cause-and-effect action playing out with real physical weight. On Aicut you pick it from the same prompt box as every other model, at a price shown before you generate.

16:9 / 9:16

4-20s

~5 min to render

Generate with Sora 2

Free account. The token price is shown before anything is generated.

Made with Sora 2

Real clips from the model, unedited.

Why creators use Sora 2

Sora 2 is at its best with one legible chain of action rather than a list of separate cool moments: a hand reaching for something, a chain reaction, an object breaking apart. It holds object and scene continuity well enough to let a prompt lean on something changing state over the shot.

Its native audio renders with the picture, so dialogue and effects land in sync without a separate voice step, and it takes shot language well when it is tied to one subject action and one camera move.

What you can make with Sora 2

•Physically grounded single-action clips: something opening, breaking, filling up or reacting, staged with real weight.
•Handheld, vlog-style and action-cam formats where natural camera movement and native audio matter.
•Multi-scene sequences up to 8 scenes, billed for the combined length of every scene at the same per-second rate rather than multiplied per scene.
•Start-frame animation, when you already have an opening image and want Sora to generate the motion from it.

How to generate with Sora 2

01
Select Sora 2
Open the AI Video Generator and pick Sora 2 from the model list, or start straight from this page.
02
Describe one clear action
Write the subject, the one action or chain of events, the camera move and the setting. One legible sequence works better than several stacked ideas.
03
Generate
The clip renders in a few minutes, at the price shown on the button before you click.
04
Post it
Download the vertical clip or publish it to TikTok, YouTube Shorts and Instagram Reels in one click.

What every Sora 2 clip includes

Native audio
Dialogue, ambience and sound effects are generated with the picture on every clip, no separate voice step.
Character consistency
A character keeps their look across the shot, useful for anything with a recurring face.
Up to 8 scenes
Multishot support strings up to 8 scenes into one sequence without leaving the tool, billed for the combined length of every scene at the same per-second rate rather than multiplied per scene.
Five length options
4, 8, 12, 16 or 20 seconds, vertical or landscape.
Start frame
Feed Sora an opening image and it generates the motion from it.

What Sora 2 does best

•Physically coherent action: one clear cause-and-effect sequence reads far better than a list of separate cool moments.
•Strong object and scene continuity within a single shot, so something changing, breaking or filling up holds up over the clip.

What to know before you generate

•Dense, multi-action prompts hurt reliability more than on some other models; one legible action per clip works best.
•Best kept to concrete physical staging rather than abstract mood-writing or fast improvised banter.

What does Sora 2 cost?

0.4 tokens per second of video ($0.24 per second on the pay-as-you-go rate), at any of this model's resolutions or lengths. A multishot sequence is billed for the combined length of every scene at this same per-second rate, never multiplied per scene.

Sora 2 FAQ

What makes Sora 2 different from other AI video models?

It is built for physically grounded action: one clear cause-and-effect sequence, with native audio rendered in the same pass as the picture rather than added afterward.

Does the price change with resolution or clip length?

Sora 2 has no resolution setting and is priced per second of video, so a longer clip costs more at the same per-second rate. A multishot sequence bills for the combined length of every scene at that rate, never multiplied per scene. The current price is shown above and on the generate button.