Skip to main content
Orisu connects to a range of hosted video models. You pick the model per node in its config panel, so the same node can run on different families depending on what you need. This page covers which models exist and what they’re good for. For what each setting does, see Model settings.

​
Generation

These families power the Generate video node, turning a text prompt and/or a first-frame image into video. Capabilities vary by family: end frames, reference images and videos, audio input, and subject references appear only on models that support them. Most families expose their quality tiers directly on the node instead of as separate models: Veo 3.1 offers Standard / Fast / Lite, Seedance 2.0 offers Standard / Fast / Mini, Kling 3.0 offers Standard / Pro / 4K / Turbo Standard / Turbo Pro, Kling O3 offers Standard / Pro / 4K, Kling 2.1 offers Standard / Pro / Master, Hailuo H3 offers H3 / H3 Max / H3 Max Turbo, Wan 3.0 offers Standard / Prime, FLUX 3 offers Standard / Draft, LTX 2.5 offers Fast / Pro, and Vidu Q3 offers Standard / Turbo. The credit estimate updates as you switch. Audio on/off is a setting (generate_audio) on models that support it, not a separate model.
Some video models generate audio and some don’t. Veo, Kling, Seedance, Wan, Hailuo H3, FLUX 3, and Gemini Omni can all produce sound, and most of them show a generate-audio toggle in the node’s settings (audio usually adds to the per-second price). Turn it off when you plan to add your own soundtrack.

​
Available generation variants

Veo (Google)

Veo 3.1 (Standard / Fast / Lite): text, a first frame, or a first and last frame to video. Veo 3.1 Reference (Standard / Fast / Lite): an 8-second clip built around up to 3 reference images

Gemini Omni (Google)

Gemini Omni Flash 1.1: text, a first and last frame, or reference images and short videos to video with audio, 360p to 4K

Kling (Kling)

Kling 3.0 (Standard / Pro / 4K / Turbo Standard / Turbo Pro), O3 (Standard / Pro / 4K), 2.6 Pro, 2.5 Turbo, 2.1 (Standard / Pro / Master), 2.0, O1, Elements (Standard / Pro), Motion Control

Hailuo (MiniMax)

Hailuo H3 (H3 / H3 Max / H3 Max Turbo), Hailuo 02 Pro, 2.3 Standard, 2.3 Fast

Seedance (ByteDance)

Seedance 2.5, 2.0 (Standard / Fast / Mini), 1.5 Pro, Pro, Pro Fast

Wan (Alibaba)

Wan 3.0 (Standard / Prime), Wan 2.7, Happy Horse 1.1

FLUX 3 (Black Forest Labs)

FLUX 3 (Standard / Draft)

LTX (Lightricks)

LTX 2.5 (Fast / Pro): text or a start and end frame to video with sound. LTX 2.5 Audio to Video (Fast / Pro): an audio track, plus an optional first frame, to a clip as long as the audio

Luma Ray (Luma AI)

Luma Ray 3.2 (text to video), Luma Ray 3.2 Image to Video

Grok Imagine Video (xAI) and Vidu

Grok Imagine Video 1.5, Grok Imagine Video, Vidu Q3 (Standard / Turbo)

​
Editing

The Edit video node rewrites existing footage from text instructions. It runs on:
  • Gemini Omni Flash 1.1: conversational edits (object replacement, style changes) without regenerating the whole clip, at 360p to 4K
  • Kling O3 (Standard / Pro / 4K) and Kling O1: rewrite a clip of 3 to 15 seconds (O1: up to 10) from instructions, guided by up to 4 reference images. Kling Motion Control copies the movement of a clip onto the character in your first frame.
  • FLUX 3: extends an existing clip that is shorter than 15 seconds
  • Grok Imagine Video 1.5: extends a clip by 6 seconds from a description of what happens next
  • Veo 3.1 Extend (Standard / Fast): adds 7 seconds, with sound, to a 720p or 1080p clip in 16:9 or 9:16
  • LTX 2.3 Reframe: re-renders a clip at 9:16, 1:1, 4:5, 5:4, or 16:9 and fills the new edges to match the footage. Instructions are optional.
  • Luma Ray 3.2: re-renders an existing clip from a prompt
  • Luma Ray 3.2 Reframe: changes a clip’s aspect ratio and fills the new space instead of cropping, for clips up to 10 seconds. Instructions are optional.
  • Grok Imagine Edit

​
Lip sync

The Generate lipsync node drives a face from an audio track. It runs on:
  • Sync Lipsync 3: re-times a video’s mouth to new audio, or makes a portrait photo talk
  • Sync Lipsync 2.0 (Quality: Standard / Pro)
  • HeyGen Lipsync (Precision / Speed): Speed costs half as much, for drafts
  • VEED Lipsync 2 and Veed Fabric
  • OmniHuman 1.5 (Standard / Turbo): turns one photo and a voice track into a talking video where the whole person moves
  • Kling

​
Upscale

The Upscale video node increases resolution. It runs on:
  • FlashVSR
  • Topaz Video Upscale (Precision / Creative / Generative): 2x or 4x, for clips up to 5 minutes
Match the model to the inputs you have. Generation models accept prompts and first frames; editing models need an existing video; lip sync needs a face plus audio; upscaling needs a finished clip.