How to make AI cat videos: 5 formats with prompts and costs

This guide compares five AI cat video formats, starting from one photo of your own cat. Each has a draft prompt, a model choice and a cost worked out from published list prices, checked on 2026-10-26.

How do you make an AI cat video?

You make an AI cat video by uploading a cat photo to an image-to-video model and writing one action and one camera move in the prompt.

1. Take a sharp, evenly lit photo of your cat. Google's Gemini docs advise high-resolution images and specific motion descriptions.
2. Write one subject, one action, one setting. "The cat sits at a kitchen table, picks up a spoon, looks at the camera."
3. Pick a model by what the format needs: audio, length or price.
4. Generate short clips. Cut them together in any editor.
5. Keep the best take and discard the rest. Plan for several takes per keeper.

Two features matter for cats. Gemini Omni Flash accepts subject references, and its docs use a cat as the example. It also outputs 9:16 and 16:9 at 720p by default. The all-in-one cat generators, such as VisionStory (15-second, 720p, 9:16 skits) and Revid (4 to 6 second scenes with automatic cat sounds), trade control for speed.

What are the main AI cat video formats?

Here are five formats you can try: a cat vlog, a cat soap opera, a singing cat, a cat doing a human job, and a dancing cat.

How do you make an AI talking cat?

An AI talking cat needs a model that makes audio, or a separate voice track laid over a silent clip.

Two methods:

• Audio in the model. Veo 3.1 and Kling 3.0 both have audio rates on Runway's API. For Veo audio problems, see Veo 3 not generating audio. Write the spoken line in quotes in the prompt and name the voice ("a low, bored voice"). The model decides how the mouth moves, and it will not always match the words.
• Voice added later. Render a silent clip, then add speech from a text-to-speech tool. Runway's price list includes ElevenLabs models at $0.01 per 50 characters for v3, so a 150-character line costs $0.03. You then line the track up with the cat's mouth in your editor.

The second method is cheaper per take and gives you control of the voice. The first skips the separate voice-generation and editing step. Google's Gemini Omni docs say voice editing is not supported and that dialogue can be generated when extending a model-generated video.

How do you make an AI cat singing video?

An AI cat singing video is a cat clip with a music or vocal track, either generated with the clip or added afterwards.

Runway lists an audio model called Seed Audio at $0.0025 per second, with a $0.05 minimum per generation, so a 10-second generation costs $0.05. The price covers the audio only.

Pick the music before you post. Music used without permission can bring copyright removal requests, strikes and blocking on YouTube. A cat singing a famous song is a copyright question, not a prompt question, so use music you made, licensed or that the platform's own library allows for your account type.

How do you make a cat soap opera or story?

A cat soap opera is a series of 4 to 6 second shots of the same cat, cut together with a script and one music bed.

The hard part is that the cat looks slightly different in each shot. Ways to reduce it:

• Use the same photo as input for every shot (see Kling image to video).
• Use a model that takes subject references, such as Gemini Omni Flash.
• Keep shots close and simple. Fewer details give the model less to change.
• Write the story as a shot list before you spend anything.

Revid's page describes 4 to 6 second scenes, which fits building a story from many short shots.

What does an AI cat video cost?

On Runway's API list prices, one 8-second Veo 3.1 clip with audio costs $3.20, and one 10-second Wan 3.0 clip at 720p costs $1.00. Runway's rate is $0.01 per credit (Runway, 2026-10-26).

The soap opera sum: 6 shots × 5 seconds × $0.13 = $3.90 for one take of each. Three takes of each shot is $11.70.

"Takes" is a planning number. Change it to match how picky you are. The AI video cost calculator does the same sum for other models. App subscriptions price differently from the API, as Runway pricing explains.

What makes an AI cat video watchable?

A watchable AI cat video is short, has one clear action, and keeps the cat's face and size steady for the whole clip.

Judge each take against this checklist:

• Does the cat look like the same cat from first frame to last?
• Is there one action, not three?
• Do the paws, whiskers and tail move in ways a cat could?
• Does the sound fit what is on screen?
• Would the clip still work with the sound off?

Discard any take that fails the first two.

Do you have to label AI cat videos, and can you monetize them?

YouTube requires disclosure when an AI cat video looks realistic or when AI-generated music is its main focus, and mass-produced, repetitive AI videos are not eligible for monetization there.

• YouTube disclosure. YouTube asks creators to disclose realistic content made with AI. Its help page says unrealistic content, such as a fantastical scene, does not need disclosure on visuals alone. A cat singing on a stage reads as fantasy. A cat sitting in a real-looking living room does not. The same page requires disclosure when AI-generated music is the main focus of the video, so a singing-cat video can need the label even when the cat is clearly fictional.
• YouTube monetization. YouTube's policy bars mass-produced, repetitive content and generic template videos without original value. A template alone does not make a video ineligible. A channel posting the same prompt with a new cat each day is the risk. Add a story, a recurring character and your own voice. More in monetizing AI videos on YouTube.
• TikTok. TikTok's synthetic media policy, announced on 2023-09-19, requires labels on AI-generated content with realistic images, audio or video.
• Likeness. Use your own cat and original characters. Get permission before using someone else's likeness, voice or protected characters.

For a dog version, see AI dog dance videos.