Seedance Kling Hailuo Grok Imagine

AI Video Models

Seedance, Kling, Hailuo and Grok Imagine on one page each: what the model is, how long its clips run, what it is good at, its limits and what a clip costs.

Open the generator See the 13 video models

Two swordsmen stand in tall grass at sunset while the camera orbits left — H3 Max Turbo · image to video · 3 s · 480P

Seedance

ByteDance. Long takes, generated sound and a lot of reference control.

ByteDance Seedance 2.5 Seedance 2.5 is ByteDance's video model for long takes. It makes clips of 4 to 30 seconds with sound from text, a first frame or up to 50 reference files, and it can also edit or extend a clip. From 137 credits for 5 s0 ByteDance Seedance 2.0 Mini ByteDance's Seedance 2.0 as an AI video generator: Ziztu runs Seedance 2.0 Mini, the faster, lower priced version, and Seedance 2.5 on a page of its own. It makes 4 to 15 second clips with sound from text, a first and last frame, or references. From 45 credits for 5 s1

Kling

Kuaishou. Sound built in, character elements, an omni model that takes references, and motion transfer.

Kuaishou Kling 3.0 Kling 3.0 is Kuaishou's video model with sound built in. It makes 3 to 15 second clips from text or a first frame, can plan a multi-shot scene in one go, and keeps characters steady with elements. From 139 credits for 5 s0 Kuaishou Kling O3 and Kling O1 Kling O3 is Kling's omni model. It takes text, a first and last frame, reference images and character elements together, makes 3 to 15 second clips with sound, and can change an existing video from a written instruction. From 139 credits for 5 s1 Kuaishou Kling Motion Control Kling Motion Control copies the movement and facial expressions from a reference video onto the character in your picture. The clip is as long as the motion video, between 3 and 30 seconds. From 156 credits for 5 s2

Hailuo

MiniMax H3. Fast, low-priced clips, a 2K model, and one that takes reference pictures.

MiniMax MiniMax H3 (Hailuo) H3 is MiniMax's open-weights video model, the successor to Hailuo 02. It makes 5 to 15 second 2K clips with stereo sound from text, frames or references, and is strong at on-screen text and following edit instructions. From 31 credits for 5 s0

Grok Imagine

xAI. Lip-synced speech, and a Lite version at a lower price.

xAI Grok Imagine Video 1.5 Grok Imagine Video 1.5 is xAI's video model with generated sound and lip-synced speech. It makes clips of 1 to 15 seconds from text, one picture or up to 7 reference images, at 480p, 720p or 1080p. From 25 credits for 5 s0

What a clip costs, by model

Credits for one 5 second clip at each model's lowest quality (dollars at the Starter plan rate of $12 for 1,000 credits)
Model	Maker	Makes	Length	5 second clip
Seedance 2.5	ByteDance	Text, Image, Reference	4 to 30 s	137 credits ($1.64), 480P
Seedance 2.0 Mini	ByteDance	Text, Image, Reference	4 to 15 s	45 credits ($0.54), 480P
Kling 3.0 Pro	Kuaishou	Text, Image	3 to 15 s	208 credits ($2.50), 1080P
Kling 3.0 Standard	Kuaishou	Text, Image	3 to 15 s	156 credits ($1.87), 720P
Kling 3.0 Turbo	Kuaishou	Text, Image	3 to 15 s	139 credits ($1.67), 720P
Kling O3	Kuaishou	Text, Image, Reference	3 to 15 s	139 credits ($1.67), 720P
Kling O1	Kuaishou	Image, Reference	3 to 10 s	139 credits ($1.67), 1080P
H3 Max Turbo	MiniMax	Text, Image	1 to 15 s	31 credits ($0.37), 480P
H3 Max	MiniMax	Text, Image	5 to 15 s	62 credits ($0.74), 480P
H3 Max References	MiniMax	Reference	1 to 15 s	62 credits ($0.74), 480P
H3 2K	MiniMax	Text, Image, Reference	5 to 15 s	161 credits ($1.93), 2K
Grok Imagine Video 1.5	xAI	Text, Image, Reference	1 to 15 s	99 credits ($1.19), 480P
Grok Imagine Video 1.5 Lite	xAI	Text, Image	1 to 15 s	25 credits ($0.30), 480P

Image models for your reference pictures

Make the cast, the place or the prop here, then use the pictures in any video model above.

Meta Meta Muse Image Muse Image is Meta's image model that works like an agent. It plans a multi-part prompt before it draws, can search the web for facts, and edits or combines up to 10 reference images at about 2.5 megapixels. From 3 credits a picture0 ByteDance Seedream 5.0 Flash Seedream 5.0 Flash is ByteDance Seed's fast, low-cost image model for making and editing pictures. Edit mode takes up to 10 reference images, and one request makes up to 6 pictures at 1K or 2K sizes. From 7 credits a picture1 OpenAI GPT Image 2.5 GPT Image 2.5 is OpenAI's image model in two tiers: Flare for fast everyday pictures and Sunburst for precise edits. People search for it as GPT Image 2 or GPT Image 2.5. Ziztu runs GPT Image 2.5. From 2 credits a picture2 Google Nano Banana Pro Nano Banana Pro is Google's image model built on Gemini 3. It makes and edits pictures at 1K, 2K or 4K, renders clear text, keeps the same people across several images, and can use web search for up-to-date facts. From 38 credits a picture3 xAI Grok Imagine Image 2.0 Grok Imagine Image 2.0 is xAI's image model for making pictures from text and editing up to 5 reference images. It offers 1k and 2k sizes, low and medium quality modes, and 13 aspect ratios. From 15 credits a picture4 Ideogram Ideogram 4.5 Ideogram 4.5 is the image model for posters, logos and accurate text in pictures. Edit mode changes part of an image and leaves the rest alone, with up to 4 reference images, an optional mask and a high-precision setting. From 8 credits a picture5

Prompts and guides

Prompts you can copy for each model, and how to turn a still into a clip.

Ziztu AI Prompts for Video and Images 101 example prompts for the AI video and image models on Ziztu, grouped by model. Each says what it makes; copy it or press Try it. Open the page Google Nano Banana Prompts 16 Nano Banana prompts for posters, character sheets, edits, products, rooms and storyboards. Each says what it makes, and each opens Nano Banana Pro with the prompt in place. From 38 credits a pictureOpen the page Google Nano Banana to Video Nano Banana makes the picture, not the video. On Ziztu Nano Banana Pro makes the still and a video model animates it, with the price of both steps shown first. Open the page

How to choose a video model

Start with what you have. With only words, text to video works on most models. With a picture you like, image to video keeps its look, and a few models take a last frame too. With a cast that must stay the same across clips, reference to video takes pictures of your characters, a place or a prop.

Then look at the numbers on each model page: how long a clip can be, which resolutions it makes, whether it makes sound, and what a clip costs. Those are facts. How a clip looks is a matter of taste, and it differs from prompt to prompt, so the AI Video Generator can run the same prompt on a second model and put the two clips side by side.

Where the facts come from

Every model page lists its sources, official pages of the maker or the platform that runs the model, with the date we read them. Prices are not copied from anyone: they are the credits Ziztu charges, computed by the same function that bills a clip. Dollars are at the Starter plan rate of $12 for 1,000 credits. Plans start at $12 a month for 1,000 credits.

Frequently asked questions

Which AI video models does Ziztu run?

13 models in the generator, in four families: Seedance, Kling, Hailuo (MiniMax H3) and Grok Imagine. Ziztu Studio also runs Kling Motion Control and the HeyGen video model. Ziztu does not run Veo, Sora, Pika or Luma.

What does an AI video cost on Ziztu?

It depends on the model, the length and the quality, and the price is shown before every run. From the models page: a 5 second clip starts at 25 credits and the table on this page lists every model. Plans start at $12 a month for 1,000 credits.

Can I try these models for free?

A new account gets 40 free credits, no card, for scripts, character art, storyboards, voices and sound. Video starts on a plan, from $12 a month. You can open the generator, pick any model and see every price without an account.

How do I compare two models?

Run the same prompt on both in the AI Video Generator. After the first clip finishes, open "Same prompt, another model", pick the second model and run it. The two clips appear side by side with their settings and prices.