AI Video Generator
What clip are we cooking today?
Turn a prompt into one clip. Set the duration and resolution, add a first frame when you need one, and choose the model that fits the shot.
Create your way.
One idea can take many directions. Explore different models, guide a clip with a frame or a reference, and refine until the motion feels right.
Text to video
Describe the subject, the motion, and the camera. Choose a duration and a resolution, then cook one clip.
Frames and references
On supported models, start from a first frame, add a last frame, or guide the clip with reference images, video, or audio.
Explore multiple models
Run the same idea on a second model and compare. Every model opens in the same prompt box, so there is nothing to relearn.
AI video generation, built for exploration.
image.cooking is an AI video generator that brings Seedance, Kling, MiniMax, Wan, HappyHorse, Gemini Omni, and Veo into one workspace. Start from a prompt, add a first frame when the model supports it, and cook one clip at a time.
Set the duration and the resolution before you start. On supported models you can add a last frame, keyframes, or reference images, video, and audio. Finished clips stay in My Creations.
ByteDance
Seedance 2.5Handles long shots and rich reference material with ease, great for finished short films and ads.
- Seedance 2.0
Detailed, high-definition visuals with synced audio, ideal for polished short videos.
- Seedance 2.0 Fast
Quicker turnaround, perfect for iterating on ideas when time is tight.
- Seedance 2.0 Mini
Lightweight and budget-friendly, great for testing ideas and batch creation.
Kuaishou
Kling 4.0Tells a longer story in one go, with stereo sound, built for narrative-driven content.
- Kling 3.0
Strong at multi-shot storytelling with picture and sound together, great for story clips and product showcases.
- Kling 3.0 Omni
Keeps characters consistent, using images, video, and voice to guide them, ideal for recurring characters.
- Kling 3.0 Turbo
Fast and great value, perfect for everyday quick videos.
MiniMax
Alibaba
Wan 3.0Text, first and last frames, and reference images all in one model, ideal for all-in-one creation.
- Wan 3.0 Prime
A faster Wan 3.0 with the same abilities, perfect when you’re short on time.
- Wan 2.7
Good at multi-shot storytelling and extending clips, ideal for continuing scenes into connected mini stories.
HappyHorse
Gemini Omni 1.1 FlashEdit video by chatting, describe a change and refine step by step, ideal for creating as you think.
- Veo 3.1
Top-tier quality and texture, great for cinematic content where polish matters.
- Veo 3.1 Fast
Balances quality and speed, ideal for everyday high-quality videos.
- Veo 3.1 Lite
An entry-level, lightweight option for low-cost quick experiments.
Which model should I use?
Not sure where to start? Pick the job, then open the model.
Model specs at a glance
Duration, resolution, and inputs for every video model, in one table.
Prices shown are for default text-to-video settings. Create always shows the exact price before you start. One run makes one clip.
How Credits work
Every generation spends Credits. The price depends on the model and your settings, such as duration and resolution. Create shows the exact price before you start, and a run only begins if your balance can cover it. Signing up with Google includes starter Credits. Email sign-up does not. Paid Kitchen plans add monthly Credits. See Pricing.
Tips for better AI video prompts
- Name the subject, the action, and the setting.
- Describe the camera: the move, the angle, and the framing.
- Say how the shot changes over time, not only how the first moment looks.
- Set the duration and the aspect before you cook. A longer clip costs more on models priced by the second.
- Add a first frame when the opening image has to match something you already have.
- Run the same prompt on two models, compare, then refine the one you prefer.
AI video generator FAQ
There is no single best model. Seedance 2.5 suits longer clips and references, Kling 4.0 suits keyframes, Seedance 2.0 Mini suits a shorter draft, and Veo 3.1 Lite is prompt only. Try the same prompt on more than one.
Yes, on models that accept a first frame. Some also take a last frame, keyframes, or reference images, video, and audio. Veo 3.1 Lite is prompt only. Create shows the inputs that model accepts.
It depends on the model. Veo 3.1 is a fixed 8 seconds. Others run from a few seconds up to 15 or 30. Create shows the durations that model accepts before you start.
One. Each run returns a single clip.
Credits are the unit image.cooking uses to measure generation costs. Create shows the exact Credit price for the model and settings you have selected before you start.
Yes. Images and videos you pay to generate include commercial use, whether you pay with a Kitchen or a Credits pack. The permission stays with the output. Starter Credits are for personal, non-commercial use. See Pricing.
Completed clips are saved to My Creations, private to your account. You can play them there and download the file.