Video

Text to video AI, with sound.

Type a scene. Watch it play, with voices, ambience and music.

Get early accessSee pricing
  • From 20 credits
  • Up to 15 s
  • Up to 1080p
  • 15 models
A woman in a yellow coat walking down a rainy city street at night
a woman in a yellow coat, rain, night

How it works.

  1. 01

    Describe the shot

    Who, what, where, and the camera. Spoken lines in quotes.

  2. 02

    Pick a model

    Then the format and length. The button shows the price.

  3. 03

    Generate

    The clip lands in your library. No watermark.

What you get.

Credits per run. You see the exact price before you start, and failed runs are refunded.

Models
15
Resolution
Up to 1080p
Length
3–15 s
Price
From 20 credits

Models you can use.

15 models, one prompt box. Each card shows its cheapest run.

All models

Prompts to try.

Copy one, or make it yours.

  • A street cook flips noodles in a flaming wok at night, steam rising, handheld close-up, the sizzle of the pan.

  • An old fisherman sits on a wooden dock at sunrise and says “the sea doesn't care how tired you are”, gulls, water lapping.

  • Vertical video: a ginger cat pushes a pen off a desk while staring at the camera, slow motion on the fall, the soft clack of the pen.

What it does.

Text to video turns a written scene into a short clip. Say who is in the shot, what happens and how the camera moves: the model renders it, and most models make the sound at the same time.

15 video models share one prompt box: Veo 3.1 for realism and dialogue, Seedance for acting, Kling for motion, lighter models for quick drafts.

  • Voices, ambience and effects made with the picture.
  • Wide, vertical or square, up to 1080p.
  • The price before you start. Failed runs refunded.

Questions.

It depends on the shot. Veo 3.1 is strong on realistic people and dialogue, Seedance 2.0 and 2.5 on acting with native sound, Kling 3.0 and 4.0 on physical motion. Here you can try them all with the same credits.

Up to 15 seconds, depending on the model. For a longer video, make several clips and join them in any editor.

On most models, yes: voices, ambience and effects, made with the picture. Kling 3.0, Wan 3.0 and PixVerse let you switch it off.

From 20 credits a clip with the lightest models. The price depends on the model, the resolution and the length, and you see it before you generate. Failed runs are refunded.

Yes, on every plan, with no watermark. Don't put real people, logos or characters you don't own in your prompts.

More tools.

All tools

Keep exploring

Tools

Use cases

Comparisons

Guides

Try Text to video.

One plan, every top model. You see the price before you start.

Get early access