MiniMax-H3-Turbo-Lora-ComfyUI

MiniMax-H3 Turbo LoRA — ComfyUI Pruned-Model Compatibility Conversions

This repository contains third-party ComfyUI compatibility conversions of the original MiniMax-H3 Turbo LoRA created by larryvrh.

The Turbo LoRA enables few-step MiniMax-H3 audio-video generation, producing joint video and synchronized audio using substantially fewer sampling steps than the standard MiniMax-H3 workflow.

This repository includes compatibility conversions of the initial Turbo weights as well as later checkpoint-500, checkpoint-850, and v4 step-600 releases.

These LoRAs are intended for the pruned/curve-form MiniMax-H3 checkpoint used by ComfyUI.

Recommended checkpoint: minimax_h3_turbo_v4_step600_ema_pruned_comfyui.safetensors

Recommended settings: 8 steps · Euler sampler · Beta scheduler

---

Workflow Examples

New Workflow Output

This is the output from the current recommended workflow using the newer v4 step-600 EMA checkpoint and recommended sampling configuration.

<video src="https://huggingface.co/drbaph/MiniMax-H3-Turbo-Lora-ComfyUI/resolve/main/assets/MiniMax_H3_audio.mp4" controls loop width="100%"> </video>

New Recommended Workflow

Download the current ComfyUI workflow

The current workflow is:

fl_minimax_h3_turbo_lora_example_workflow.json

Recommended starting configuration:

Checkpoint: minimax_h3_turbo_v4_step600_ema_pruned_comfyui.safetensors
Steps: 8
Sampler: euler
Scheduler: beta
LoRA strength: 1.0

---

Recommended Setup

For the best overall results in this repository, start with:

minimax_h3_turbo_v4_step600_ema_pruned_comfyui.safetensors

Recommended ComfyUI settings

The v4 step-600 EMA checkpoint is currently the recommended general-purpose version in this repository.

For most generations, 8 steps with Euler and the Beta scheduler is the recommended starting configuration for these ComfyUI compatibility versions.

---

Old Workflow Output

The following video and workflow are retained as an example of the previous Turbo LoRA configuration.

<video src="https://huggingface.co/drbaph/MiniMax-H3-Turbo-Lora-ComfyUI/resolve/main/assets/MiniMax_H3_00060-audio.mp4" controls loop width="100%"> </video>

Old Workflow

Download the old ComfyUI workflow

The previous workflow has been renamed to:

fl_minimax_h3_turbo_lora_example_workflow_old.json

---

Original Project

The original LoRA weights, training work, distillation method, audio/video sampling implementation, documentation and source files were created and released by larryvrh.

Original repository: larryvrh/MiniMax-H3-Turbo-Lora

I did not train, distill or create the original Turbo LoRA weights.

This repository provides modified compatibility versions intended to allow the compatible LoRA adapters to load with ComfyUI's built-in MiniMax-H3 LoRA loader when using the pruned/curve-form model.

Full credit for the original MiniMax-H3 Turbo LoRA, training and distillation method belongs to larryvrh.

---

Included Files

The EMA and non-EMA files remain separate and contain different tensor values.

The files without checkpoint identifiers preserve the earlier preview weights.

The ckpt500 and ckpt850 files belong to the older training line, while the v4_step600 files use the newer v4 training recipe.

These files are partial compatibility versions of the original full-model Turbo LoRAs for use with the pruned/curve-form MiniMax-H3 architecture.

---

Which File Should I Use?

Recommended — v4 Step-600 EMA

Start with:

minimax_h3_turbo_v4_step600_ema_pruned_comfyui.safetensors

Recommended configuration:

• Steps: 8
• Sampler: euler
• Scheduler: beta
• LoRA strength: start at 1.0
• Video sigma shift: 12
• Audio sigma shift: 4-6

This is the recommended checkpoint for most generations.

v4 Step-600 non-EMA

For comparison:

minimax_h3_turbo_v4_step600_pruned_comfyui.safetensors

The non-EMA weights are provided primarily for testing and comparison against the recommended EMA checkpoint.

Checkpoint-850

EMA:

minimax_h3_turbo_4step_ema_ckpt850_pruned_comfyui.safetensors

Non-EMA:

minimax_h3_turbo_4step_ckpt850_pruned_comfyui.safetensors

These are retained for comparison with the newer v4 training line.

Checkpoint-500

EMA:

minimax_h3_turbo_4step_ema_ckpt500_pruned_comfyui.safetensors

Non-EMA:

minimax_h3_turbo_4step_ckpt500_pruned_comfyui.safetensors

These preserve the earlier further-trained checkpoint-500 release.

Initial Preview Weights

The following files preserve the initial preview versions:

• minimax_h3_turbo_4step_pruned_comfyui.safetensors
• minimax_h3_turbo_4step_ema_pruned_comfyui.safetensors

They are kept primarily for comparison and reproducibility.

---

Recommended Workflow

The current recommended workflow file is:

fl_minimax_h3_turbo_lora_example_workflow.json

Download:

fl_minimax_h3_turbo_lora_example_workflow.json

It is intended to be used with:

minimax_h3_turbo_v4_step600_ema_pruned_comfyui.safetensors

Recommended configuration:

Steps: 8
Sampler: euler
Scheduler: beta
LoRA strength: 1.0
Video sigma shift: 12
Audio sigma shift: 4-6

The corresponding example output is:

assets/MiniMax_H3_audio.mp4

View / download MiniMax_H3_audio.mp4

---

Previous Workflow

The previous example workflow is retained as:

fl_minimax_h3_turbo_lora_example_workflow_old.json

Download:

fl_minimax_h3_turbo_lora_example_workflow_old.json

Its corresponding example output is:

assets/MiniMax_H3_00060-audio.mp4

View / download MiniMax_H3_00060-audio.mp4

---

About the Original Turbo LoRA

The MiniMax-H3 Turbo LoRA is designed to generate joint video and synchronized stereo audio using dramatically fewer sampling steps than the standard MiniMax-H3 generation process.

The original project demonstrated generation using substantially fewer sampling steps than the normal MiniMax-H3 workflow, providing a large reduction in sampling time.

The newer Turbo checkpoints continue to improve upon the initial preview release.

Refer to the original repository for the author's current documentation, training updates, standalone generation implementation and sampling recommendations:

larryvrh/MiniMax-H3-Turbo-Lora

---

Why This Conversion Was Made

The original Turbo LoRA files use tensor keys such as:

blocks.0.attn.qkv_proj.lora_A.weight

ComfyUI's built-in MiniMax-H3 LoRA resolver expects keys under the model namespace:

diffusion_model.blocks.0.attn.qkv_proj.lora_A.weight

Without that namespace, ComfyUI may report warnings such as:

[WARNING] lora key not loaded: blocks.0.attn.out_proj.lora_A.weight

The following namespace conversion was applied to the retained tensors:

• blocks.* → diffusion_model.blocks.*
• token_refiner.* → diffusion_model.token_refiner.*

The retained tensor values, BF16 dtype, tensor shapes, LoRA rank and A/B orientation were preserved.

---

Conversion Details

Each original full-model Turbo LoRA contained:

Each pruned-model compatibility conversion contains:

The pruned adapters were:

• 50 AdaLN projection pairs from blocks.0 through blocks.49;
• 1 AdaLN projection pair from final_layer.adaln_proj.linear.

The retained adapters include:

• all main-block attention adapters;
• all main-block MLP adapters;
• all token-refiner attention adapters;
• all token-refiner MLP adapters.

The retained attention, MLP and token-refiner tensors have dimensions compatible with the pruned/curve-form MiniMax-H3 checkpoint.

---

Validation

The converted LoRA structure was checked for:

• complete LoRA A/B adapter pairs;
• no orphan A or B tensors;
• retained keys using the diffusion_model.* namespace;
• no duplicated diffusion_model.diffusion_model.* prefixes;
• no incompatible adaln_proj tensors;
• preserved BF16 dtype;
• preserved tensor shapes;
• preserved LoRA ranks;
• separate EMA and non-EMA weights;
• no unresolved retained keys during ComfyUI loader-level validation.

The original source files were not overwritten.

---

Optional Acceleration

The LoRAs have been tested working in ComfyUI with optional acceleration methods including:

• SageAttention
• Sol Attention
• Gradient
• Spectrum

These acceleration methods are optional and are not included in this repository.

They may also be combined with the Turbo LoRA depending on your ComfyUI workflow and available custom nodes.

---

Important Limitation

These files are partial ComfyUI compatibility conversions intended specifically for the pruned/curve-form MiniMax-H3 architecture.

The original full-model Turbo LoRAs contain AdaLN adapters that do not directly match the corresponding dimensions of the pruned/curve-form MiniMax-H3 checkpoint.

Those incompatible AdaLN tensors are therefore excluded from these compatibility versions.

The retained attention, MLP and token-refiner LoRA tensors remain compatible with the pruned architecture.

Because these are partial compatibility conversions, they should not be assumed to be numerically identical in behaviour to using the original full Turbo LoRA with its intended full MiniMax-H3 base.

---

Usage

Place the LoRA files inside:

ComfyUI/models/loras/

Use them with the pruned/curve-form MiniMax-H3 checkpoint for ComfyUI.

Recommended LoRA

minimax_h3_turbo_v4_step600_ema_pruned_comfyui.safetensors

Recommended settings

• Steps: 8
• Sampler: euler
• Scheduler: beta
• Video sigma shift: 12
• Audio sigma shift: 4-6
• LoRA strength: 1.0 recommended starting point

The recommended workflow for these settings is:

Download fl_minimax_h3_turbo_lora_example_workflow.json

The corresponding example output is:

View assets/MiniMax_H3_audio.mp4

If necessary, LoRA strength can be adjusted slightly for individual generations.

A reasonable starting range is approximately:

0.8-1.2

When comparing checkpoints, use the same:

• prompt;
• seed;
• resolution;
• frame count;
• base model;
• sampler;
• scheduler;
• LoRA strength;
• workflow.

This gives a meaningful comparison between checkpoint versions.

---

Sampling Step Guidance

The Turbo LoRA can operate at extremely low step counts, but image quality and motion consistency generally improve when moving beyond the minimum.

4 steps

Maximum speed.

Useful for testing and rapid previews.

6 steps

Good speed/quality compromise.

8 steps

Recommended for the v4 step-600 EMA compatibility conversion.

Use:

8 steps
Euler sampler
Beta scheduler

This is the recommended configuration used by the current example workflow.

---

Audio and Scheduler Notes

MiniMax-H3 generates video and synchronized audio as separate streams with different scheduling requirements.

Incorrect audio scheduling can result in:

• blown-out audio;
• severe distortion;
• noise-like audio;
• unstable audio energy;
• completely broken audio.

For the current recommended ComfyUI compatibility workflow in this repository:

LoRA: minimax_h3_turbo_v4_step600_ema_pruned_comfyui.safetensors
Steps: 8
Sampler: euler
Scheduler: beta
Video sigma shift: 12
Audio sigma shift: 4-6
LoRA strength: 1.0

Do not automatically assume broken audio is caused by the LoRA.

Check:

• sampler configuration;
• scheduler configuration;
• audio sigma shift;
• video sigma shift;
• ComfyUI version;
• MiniMax-H3 sampling implementation;
• custom sampler nodes.

---

Original Turbo Custom Node

The original Turbo author also provides a dedicated ComfyUI custom node:

GitHub — Larryvrh/ComfyUI-MiniMax-H3-Turbo

The node includes a MiniMax-H3 Turbo LoRA loader and dedicated sampling functionality designed for MiniMax-H3's video/audio scheduling behaviour.

A workflow is also included in the custom-node repository.

The custom node evolves alongside the upstream Turbo weights, so users choosing that implementation should follow the settings documented in its current README.

The settings documented in this repository refer to the tested pruned-model compatibility workflow, where the recommended starting point is:

v4 step-600 EMA · 8 steps · Euler · Beta scheduler

---

Compatibility

These LoRAs are intended for the pruned/curve-form MiniMax-H3 checkpoint used by ComfyUI.

The compatibility conversions are particularly useful when the original full-model LoRA cannot be loaded directly because of incompatible AdaLN dimensions.

Optional acceleration methods confirmed working with these conversions include:

• SageAttention
• Sol Attention
• Gradient
• Spectrum

Accelerators can generally be stacked where the corresponding ComfyUI implementations support it.

---

Naming Notes

The upstream project contains multiple generations of Turbo training.

Older naming

Files such as:

minimax_h3_turbo_4step_ema_ckpt500.safetensors

and:

minimax_h3_turbo_4step_ema_ckpt850.safetensors

belong to the older training line.

In these filenames, 4step refers to the few-step Turbo sampling target.

Newer v4 naming

Files such as:

minimax_h3_turbo_v4_step600_ema.safetensors

use the newer naming convention.

Here:

• v4 refers to the newer training recipe;
• step600 refers to training step 600;
• ema refers to the exponential-moving-average weights.

The corresponding pruned ComfyUI compatibility conversion in this repository is:

minimax_h3_turbo_v4_step600_ema_pruned_comfyui.safetensors

This is the recommended checkpoint.

---

Repository Workflow Files

Current workflow

fl_minimax_h3_turbo_lora_example_workflow.json

Download current workflow

Current workflow output:

assets/MiniMax_H3_audio.mp4

View current workflow output

Previous workflow

fl_minimax_h3_turbo_lora_example_workflow_old.json

Download previous workflow

Previous workflow output:

assets/MiniMax_H3_00060-audio.mp4

View previous workflow output

---

Attribution

Original MiniMax-H3 Turbo LoRA author:

larryvrh

Original project:

MiniMax-H3-Turbo-Lora

Original ComfyUI Turbo custom node:

ComfyUI-MiniMax-H3-Turbo

This repository contains third-party, partial ComfyUI compatibility versions for the pruned/curve-form MiniMax-H3 checkpoint, including:

• initial Turbo preview weights;
• checkpoint-500 variants;
• checkpoint-850 variants;
• v4 step-600 variants.

I do not claim authorship of:

• the original Turbo LoRA weights;
• the original LoRA training;
• the original distillation method;
• the original few-step sampling method;
• the original dual video/audio scheduling implementation;
• the original generation scripts;
• the original documentation;
• the MiniMax-H3 base model;
• the pruned/curve-form MiniMax-H3 checkpoint.

The modifications in this repository are compatibility conversions intended to make the compatible portions of the upstream Turbo LoRAs usable with the pruned/curve-form MiniMax-H3 architecture in ComfyUI.

Please retain attribution to larryvrh and comply with the licences and terms of the original LoRA and MiniMax-H3 base model when using or redistributing these files.