Skip to content
Docs
Docs

Models

The image, video and audio models the Assistant can use, and which it chooses for what you ask.

The Assistant makes and changes images, and makes videos, with these models. Models marked Availableare offered in Otherspace now. Evaluating models are being tested, and are not offered yet. Experimental models are offered in the app with an Experimental badge while they are being tested.

Auto

Unless you choose a model, the Assistant picks one for what you ask (Auto, shown with the model it will use). Choose a model in the Assistant's model menu to use it instead. A model that can't take your prompt (too many images, say) says why and can't be chosen.

You ask forAuto uses
A new imageNano Banana 2
A change to one selected imageNano Banana 2.1
Iterate on an image (right-click → Iterate)FLUX.3 Klein
A new image from several selected imagesNano Banana 2.1
A video from wordsMiniMax H3 Turbo
A video from one selected imageMiniMax H3 Turbo
A video guided by several selected imagesMiniMax H3 Turbo
A character from an image, moving as a video you chose movesKling 3 Motion Control
Upscale, in the Image EditorTopaz Wonder 3.5
Fill new areas, in the Image EditorBria Expand
Split layers (High Quality), in the Image EditorSeedream 5 Flash Layerize

Images are made at 1K. Videos are made at 768p and 5 seconds long. Every image or video a model makes is kept with its prompt, the model and what it was made from.

Image models

Video models

Seedance 2.0 Fast

ByteDance

Available

Fast video

Images in
Up to 9
Output
480p to 720p, 4 to 15 s, with sound

Seedance 2.0

ByteDance

Available

Quality video

Images in
Up to 9
Output
480p to 1080p, 4 to 15 s, with sound

MiniMax H3 Turbo

MiniMax

Available

Fastest, cheapest video

Images in
Up to 9
Output
480p to 1080p, 1 to 15 s, with sound

MiniMax H3 Max

MiniMax

Available

Quality, fast video

Images in
Up to 9
Output
480p to 1080p, 1 to 15 s, with sound

MiniMax H3 Lite

MiniMax

AvailableExperimental

Lighter video

Images in
Up to 9
Output
480p to 768p, 4 to 15 s, with sound

MiniMax H3 Highspeed

MiniMax

AvailableExperimental

High-speed video

Images in
Up to 9
Output
768p, 4 to 15 s, with sound

Wan 3

Alibaba

Available

Long videos

Images in
Up to 10
Output
480p to 1080p, 2 to 30 s, with sound

Kling 3

Kuaishou

Available

Cinematic video

Images in
Up to 2
Output
720p, 3 to 15 s, with sound

Kling 3 Pro

Kuaishou

Available

Cinematic video, 1080p

Images in
Up to 2
Output
1080p, 3 to 15 s, with sound

Topaz Video Upscale

Topaz Labs

Available

Faithful video upscaling

Images in
Up to 1
Output
720p to 2160p, 1 to 60 s

VEED Background Removal

VEED

Available

Cut-out video

Images in
Up to 1
Output
480p to 2160p, 1 to 60 s

FLUX 3 Video Edit Fast

Black Forest Labs

Available

Targeted video edits

Images in
Up to 1
Output
480p to 720p, 1 to 14 s

Kling O1 Edit

Kuaishou

Available

Video editing

Images in
Up to 1
Output
720p to 2160p, 3 to 10 s

Kling 3 Motion Control

Kuaishou

Available

A video’s motion, on a character

Images in
Up to 2
Output
720p, 3 to 10 s, with sound

Kling 3 Motion Control Pro

Kuaishou

Available

A video’s motion, on a character, 1080p

Images in
Up to 2
Output
1080p, 3 to 10 s, with sound

Audio models

Voice-overs, sound effects and music, each made from your words.