Workflow asset library

MiniMax H3 ComfyUI Workflow JSON Library

MiniMax H3 supports several local workflow paths depending on the input and control you need. This library records ready-to-import ComfyUI JSON workflows for text-to-video, image-to-video and multimodal references, while clearly marking assets that are not currently verified in this repository.

Choose the workflow based on the material you already have rather than downloading every graph. For full installation, model placement, VRAM requirements and troubleshooting, use the MiniMax H3 ComfyUI installation guide.

Library type
Downloadable ComfyUI workflows
Model family
MiniMax H3
Last workflow review
September 15, 2026
Full installation guide
MiniMax H3 ComfyUI Guide

Where Can I Download MiniMax H3 ComfyUI Workflows?

You can download ready-to-import MiniMax H3 ComfyUI JSON workflows from this library for T2V, I2V and reference-to-video. FL2VA and Turbo records remain visible but are not presented as downloads until a distinct, verified JSON asset is registered.

Which MiniMax H3 Workflow Should You Download?

Starting materialRecommended workflowBest for
Text prompt onlyT2VGenerating a scene from text
One starting imageI2VAnimating a still image
First + last frameFL2VAControlling start and end states
Multiple referencesR2V / Ref2VAMultimodal reference control
Existing optimized setupTurboExperimental community inference

If you only have a text prompt, start with T2V. If you already have an image, use I2V. Use FL2VA when both the opening and ending state matter, and use Ref2VA/R2V when multiple image, video or audio references need to guide the result.

MiniMax H3 Workflow Asset Records

T2V Workflow

Use when:
You want to generate from text without a reference image.
Workflow family
FL2VA
Input
Text
Output
H3 audiovisual video
JSON status
Available
Source class
Comfy-Org template
Required model family
MiniMax H3 FL2VA
Workflow version
Upstream-managed
Last JSON update
Not recorded locally
Last tested ComfyUI build
Not recorded
Known incompatible nodes
None recorded
Known issue
Upstream templates can change; review loader selections after import.

I2V Workflow

Use when:
You have one starting image to animate.
Workflow family
FL2VA
Input
Image + text
Output
H3 audiovisual video
JSON status
Available
Source class
Comfy-Org template
Required model family
MiniMax H3 FL2VA
Workflow version
Upstream-managed
Last JSON update
Not recorded locally
Last tested ComfyUI build
Not recorded
Known incompatible nodes
None recorded
Known issue
Confirm the imported graph exposes the image inputs you need.

FL2VA Workflow

Use when:
Both the opening and ending frame need to control the result.
Workflow family
FL2VA
Input
First + last frame + text
Output
H3 audiovisual video
JSON status
Not available in this library
Source class
No separate local asset recorded
Required model family
MiniMax H3 FL2VA
Workflow version
Not recorded
Last JSON update
Not recorded
Last tested ComfyUI build
Not recorded
Known incompatible nodes
None recorded
Known issue
Use a current compatible I2V/FL2VA graph and add the last-frame input.
JSON not registeredView Technical Setup Guide

R2V / Ref2VA Workflow

Use when:
Multiple image, video or audio references guide the result.
Workflow family
Ref2VA
Input
Image/video/audio references + text
Output
H3 audiovisual video
JSON status
Available
Source class
Comfy-Org template
Required model family
MiniMax H3 Ref2VA
Workflow version
Upstream-managed
Last JSON update
Not recorded locally
Last tested ComfyUI build
Not recorded
Known incompatible nodes
None recorded
Known issue
Reference media limits and node inputs may change with ComfyUI builds.

Turbo Workflow

Use when:
You already have a compatible optimized community setup.
Workflow family
Community
Input
Depends on the community build
Output
H3 audiovisual video
JSON status
Not available in this library
Source class
Community / experimental
Required model family
Build-specific H3 Base + compatible acceleration asset
Workflow version
Not recorded
Last JSON update
Not recorded
Last tested ComfyUI build
Not recorded
Known incompatible nodes
None recorded
Known issue
Compatibility and quality depend on the selected community build.
JSON not registeredView Technical Setup Guide

MiniMax H3 Workflow Registry

WorkflowFamilyInputSource classJSONLast tested
T2VFL2VATextComfy-Org templateDownloadNot recorded
I2VFL2VAImage + textComfy-Org templateDownloadNot recorded
FL2VAFL2VAFirst + last frame + textNo separate local asset recordedNot registeredNot recorded
R2V / Ref2VARef2VAImage/video/audio references + textComfy-Org templateDownloadNot recorded
TurboCommunityDepends on the community buildCommunity / experimentalNot registeredNot recorded

How to Import a MiniMax H3 Workflow Into ComfyUI

  1. Download the JSON file for the workflow you need.
  2. Open ComfyUI.
  3. Drag the JSON file onto the canvas or use Load.
  4. Point the model loaders to your installed H3 model files.
  5. Confirm no required nodes are missing.
  6. Enter your prompt or reference inputs and queue the workflow.

Need help installing MiniMax H3, placing model files, resolving missing nodes or fixing VRAM errors? Use the MiniMax H3 ComfyUI installation guide.

Hardware Requirements

MiniMax H3 hardware needs vary by model format, resolution, workflow, GPU, system RAM and offloading strategy.

MiniMax H3 local hardware requirements

Workflow Won't Load?

Update ComfyUI, confirm the file downloaded correctly, check required nodes, and verify that the correct H3 model family is available.

MiniMax H3 ComfyUI installation guide

Related H3 documentation