Ovi

Ovi is an experimental Veo-3–style video+audio generation model that takes text and/or image inputs and produces 5-second, 24 FPS clips with synchronized audio. It supports structured prompting for dialogue and sound effects via `<S>...</S>` (speech) and `<AUDCAP>...</AUDCAP>` (background audio) tags.

Use this model

Speed

Price

Inputs

TextImage

Outputs

VideoAudio

Lab

Character AI

Ovi sample outputs

Captured directly from our eval suite. Click any tile to inspect the full render.

Action sequence test

“A hooverboarder performing an ollie trick in a city skate park, smooth camera following motion, golden hour lighting”

EngineId: ovi

•

Seed: 65505

Comic book history style test

“A 4-panel comic book style history of the world, from the beginning of time to the present day, with a focus on the most important events and figures in history.”

EngineId: ovi

•

Seed: 2999