ChatCut Models
AI video & image models
Seedance, Kling, GPT Image and more, in one editor. Pick a model, generate the shot, keep editing.
Video models
Generate a clip, then cut it on the same timeline.
-
ByteDance
Seedance 2.5
Mix up to 30 image, 10 video and 10 audio references, or edit and extend a clip you already have.
Explore -
ByteDance
Seedance 2.5 Flash
A faster, cheaper Seedance 2.5 for trying shots at 480p or 720p, about a third less than the full model at 720p.
Explore -
ByteDance
Seedance 2.5 4K
Native 4K video with 10-bit color, for ads, film and big screens. Draft at 480p, then finish in 4K.
Explore -
Kuaishou
Kling 4.0
The longest single take in ChatCut: up to 30 seconds with native sound, and 21:9 output.
Explore -
Kuaishou
Kling 4.0 Flash
Kling 4.0's inputs and native sound for 40% less, at 720p and up to 20 seconds. For trying the move before the final.
Explore -
MiniMax
MiniMax H3
H3 and the faster H3 Max: 4–15 second clips with sound, from text, frames or image, video and audio references, up to 2K.
Explore -
Google
Gemini Omni 1.1 Flash
Edit or extend a clip you already have, or make a new one from text, frames or up to three images. Cheap 360p previews.
Explore -
ByteDance
Seedance 2.0
Text, first and last frames or mixed references, with native 4K output on clips of up to 15 seconds.
Explore -
Kuaishou
Kling 3.0
Text, first and last frames and reference images in one request, with native audio. Paused for new generations for now.
Explore
Image models
Make stills, posters and first frames for your video.
-
OpenAI
GPT Image 2.6
Posters, product shots and transparent stickers in 10 ratios. Edit your own photos, then animate them.
Explore -
Google
Nano Banana
Pro, 2.1 and 2 in one menu: readable text, product shots from your photos, and banners from 8:1 to 1:8.
Explore -
Kuaishou
Kling Image 4.0
Images up to 8K, a set of matching shots from one prompt, or a full product page from one photo.
Explore -
ByteDance
Seedream 5.5
Edit one image round after round while the rest of the picture stays put.
Explore
Which model should I pick?
Start from the shot you need. The Model Capability Matrix has every input limit, length and resolution.
- One long, continuous shot Kling 4.0 Up to 30 seconds in a single take, with sound.
- Change or extend a clip you already have Seedance 2.5 Edit and Extend modes take your source video directly.
- A quick edit to a short clip you shot Gemini Omni 1.1 Flash Describe the change and it keeps the people and the camera move; clips up to 10 seconds.
- Reference images, video and audio in one request, up to 2K MiniMax H3 Up to 12 reference files, or a last frame on its own; 4–15 seconds with sound.
- Native 4K output Seedance 2.5 4K Native 4K with 10-bit color. Seedance 2.0 also renders 4K, up to 15 seconds.
- Many quick takes before the final Seedance 2.5 Flash About a third cheaper than Seedance 2.5 at 720p. For one shot you plan to keep, a Seedance 2.5 480p draft can become the 1080p or 4K final.
- A still, poster or sticker GPT Image 2.6 Then drag the image into a video model as a first frame.
- One image, edited round after round Seedream 5.5 Change one detail per round and the rest of the picture stays put.
- Readable text or a product that must match your photo Nano Banana Pro Up to 14 reference images, and small text that comes out spelled right.
- Lots of cheap image drafts, or a very wide banner Nano Banana 2.1 About 0.2 credits an image, with 8:1, 4:1, 1:4 and 1:8 ratios.
- A print-size image, or several matching shots of one product Kling Image 4.0 The only image model in ChatCut with 8K, plus sets of 2–10 related images in one job.
Guides, prompts and comparisons
Step-by-step posts from the ChatCut blog for every model on this page.
Seedream
Seedream 5.5GPT Image
GPT Image 2.6MiniMax
MiniMax H3 & H3 MaxQuestions about AI models in ChatCut
Which AI models can I use in ChatCut?
Video Gen offers Seedance 2.5 (up to 4K) and Seedance 2.5 Flash, Seedance 2.0 (plus its Fast and mini versions), Kling 4.0, Gemini Omni 1.1 Flash, MiniMax H3 and H3 Max. For images, there are Seedream 5.5, GPT Image 2.6, Nano Banana Pro, 2.1 and 2, and Kling Image 4.0; the image generation docs list the older image models in the selector. Kling 3.0 is temporarily paused for new generations. The Model Capability Matrix in the docs lists the current inputs, lengths and resolutions for each.
Do I need a separate account for each model?
No. Every model runs inside ChatCut with your ChatCut credits, so you pick a model in the AI panel instead of signing up for each provider. Generation uses credits, and most models need a paid plan; the credit estimate appears beside Send before you generate.
How do I choose between them?
Start from the shot you need: length, whether you have a source clip to edit, the resolution you deliver at, and how many takes you expect. The guide above covers the common cases, and each model page shows real examples.
Can I edit what a model generates?
Yes. Every result lands in My Assets and drops onto the same timeline as the rest of your project, so you can trim it, caption it, add music or a voiceover, and export without switching tools.














