AI Video Models: List & Comparison

AI video models differ by developer and by what they are best at — photorealism, anime-style output, natural motion, or speed. This page lists the main models with their developer, strengths and whether you can use them on Sesshu.

Last updated:

Video generation models

Switch between the video generation models in one workspace. Pick by strength — the credit table below has the cost.

Models you can use on Sesshu right now

Credit costs are calculated automatically from the settings in src/config/pricing-user.ts and vary with resolution, duration and audio. Veo 3.1 Lite, Grok Imagine, MiniMax H3, Gemini Omni Flash and Vidu Q4 always generate audio; Hailuo 02 is image-to-video only.
ModelCredits per video (approx.)Text to videoImage to video
Seedance 1.0 Pro Fast30 credits (720p, 5s, no audio)YesYes
MiniMax H3 Turbo (paid credits only)50 credits (768p, 5s, with audio)YesYes
Hailuo 0254 credits (768p, 6s, no audio)—Yes
Seedance 2.0 Mini60 credits (720p, 5s, with audio)YesYes
MiniMax H360 credits (768p, 5s, with audio)YesYes
Seedance 1.5 Pro70 credits (720p, 5s, with audio)YesYes
Veo 3.1 Lite70 credits (720p, 6s, with audio)YesYes
Grok Imagine78 credits (720p, 6s, with audio)YesYes
Vidu Q4 (paid credits only)105 credits (720p, 5s, with audio)—Yes
Wan 3.0120 credits (720p, 5s, with audio)YesYes
Kling 3.0155 credits (720p, 5s, no audio)YesYes
Gemini Omni Flash155 credits (720p, 6s, with audio)YesYes
Seedance 2.0 Fast170 credits (720p, 5s, with audio)YesYes
Wan 3.0 Prime170 credits (720p, 5s, with audio)YesYes
Kling 4.0240 credits (720p, 5s, with audio)YesYes
Seedance 2.5 Flash (paid credits only)275 credits (720p, 5s, with audio)YesYes
Seedance 2.0280 credits (720p, 5s, with audio)YesYes
FLUX 3 Max (paid credits only)280 credits (720p, 5s, with audio)YesYes
Seedance 2.5 (paid credits only)420 credits (720p, 5s, with audio)YesYes

Free credits come from the daily check-in — sign in every day and claim yours (resets every Monday UTC).

How the main AI video models compare

Model names are trademarks or registered trademarks of their respective owners. Strengths summarise general assessments and vary by version.
ModelStrengthsOn Sesshu
Seedance (2.5)Known for multi-cut composition and natural motion. Up to 30s, synced audio, 480p–1080p (4K rolling out officially)Supported (including 2.5)
Seedance 2.5 FlashFast, lower-cost Seedance 2.5 tier. 480p/720p, 4–30s, synced audio, multimodal referencesSupported
Seedance 2.0On Sesshu: 480p / 720p / 1080p, 4–15s, with or without audio (same credits)Supported
Seedance 2.0 FastThe faster version of 2.0. On Sesshu: 480p / 720p, 4–15sSupported
Seedance 2.0 MiniThe lower-cost version of 2.0. On Sesshu: 480p / 720p, 4–15sSupported
Kling 3.0Strong natural motion and high resolution. On Sesshu: 720p / 1080p, 3–15s, with or without audioSupported
Kling 4.0Built for long, high-resolution generation. 4.0 was released on October 15, 2026 (up to ~30s, 4K, Omni Reference)Supported
Vidu (Q4)Strong at anime and illustration-style output and character consistency. On Sesshu: image to video (720p / 1080p / 2K / 4K, 3–16s, with audio)Supported
FLUX 3 MaxThe video model of the FLUX series. On Sesshu: 720p / 1080p, 5–20s, with or without audioSupported
WanUp to Wan 2.2 the weights are open and it runs locally. The latest Wan 3.0 is cloud service-onlySupported
GenjutsuVideo-to-video tools (Motion Transfer / Object Swap / Restyle). The underlying model is not disclosedNot supported (explainer only)
Grok ImagineGenerates video with audio, from 6 seconds (up to 720p on Sesshu)Supported
MiniMax H3A recent model with native audio. On Sesshu: 768p, 4–15s (always with audio)Supported
Hailuo 02Strong at facial expression and motion. On Sesshu: image to video (512p / 768p, 6 / 10s)Supported (image to video only)
Gemini Omni FlashGemini Omni 1.1 Flash. Text, image or start/end frames to video. On Sesshu: 720p / 1080p / 4K, 4 / 6 / 8 / 10sSupported
Veo 3.1 LiteGenerates video and audio together. On Sesshu: 720p, 4 / 6 / 8s (always with audio)Supported
SoraThe app was discontinued on April 26, 2026Discontinued

How to choose a model

  • Just want to try it for free: Start with the lowest-credit models to practise prompting.
  • Product videos and ads: Prioritise image quality and stable motion, and pick a higher-tier model.
  • Anime and illustrations: Pick a model that is strong at image-to-video.
  • Kling 4.0 in detail: Released October 15, 2026 — specs and new features are in Kling 4.0 explained.
  • Vidu Q4 status: What is confirmed and what has not been announced is in Vidu Q4 explained.
  • See what the models can do: Example videos with the prompts that produced them are in the showcase gallery.

FAQ

When will new models become available?

+

We add new models once their quality and cost have been verified. Supported models are announced on this page.

Do different models cost different amounts?

+

Yes. Credit costs vary by model, video length and resolution. You can see the exact cost in the generator before you generate.

Which models can I try with free credits?

+

All of the models above except those marked "paid credits only" (such as Seedance 2.5). Starting with a low-credit model is a good idea.