🟣 Deploy on Runpod

🟡 Deploy on Vast.ai

ComfyUI-QwenVL-Mod — Enhanced Vision-Language with MiniMax H3 Version 2.8.0 (2026/09/02) — 🎬 MiniMax H3 Native Video+Audio + NVFP4 Blackwell + SOL-ATTN + Spectrum (native only) + lightx2v Turbo LoRA + Wildcards + 10Eros-Max Support + Camera Tag Dropdown + Unified FL2VA Loop

⬆️ 2026/09/04 UPDATE ⬆️

🔧 Turbo LoRA Switch — larryvrh → lightx2v 8-step 768p

All Turbo workflows now use lightx2v Turbo LoRA 8-step 768p (Apache-2.0) instead of larryvrh v4-600:

FL2VA: minimax_h3_fl2v_turbo_8step_v1.0_768p_comfyui_bf16.safetensors (~1.96 GB)
Ref2VA: minimax_h3_ref2v_turbo_8step_v1.0_768p_comfyui_bf16.safetensors (~1.96 GB)
No custom node required — standard ComfyUI LoRA loader works
Removed Larryvrh/ComfyUI-MiniMax-H3-Turbo custom node
Trained at 1344×768 — matches our native resolution exactly

🚫 Spectrum + Sol-Attn Bypass with Turbo — Verified

Community testing confirmed that Spectrum and Sol-Attn cause fallbacks and slowdown when used with Turbo LoRA at 8 steps:

Spectrum: forecasts steps that don't have enough context at 8-step — fallbacks double the work. "Near-useless at 10 steps" (community)
Sol-Attn (tau 1.3→0.8): aggressive sparse approximation on 8-step trajectory causes reconstruction fallbacks
Rule: bypass both Spectrum and Sol-Attn when Turbo LoRA is active. They are for 20-step native only

Turbo pure: Turbo LoRA ON · Spectrum OFF · Sol-Attn OFF · euler + simple · 8 steps Native + Spectrum: Turbo LoRA OFF · Spectrum ON · Sol-Attn ON (tau 1.0) · res_multistep + simple · 20 steps

🔄 10Eros-Max Switch — DmitryDB → cicalooo TURBO Hybrid Beta3

The 10Eros-Max model has been switched to cicalooo's ComfyUI-native INT8 ConvRot Turbo Hybrid Beta3:

Model: 10Eros_Max_h3_TURBO-hybrid_beta3_int8_convrot_skip_edges.safetensors (~22.5 GB)
TURBO fused in checkpoint — no separate t8star LoRA needed
Boundary blocks 0, 1, 48, 49 left in BF16 for stability
Removed DmitryDB 10Eros model + t8star compatibility LoRA

⚡ Comfy Kitchen Attention — Replaces SageAttention

All MMH3 Docker/provisioning now uses --use-ck-attention (Comfy Kitchen Attention) instead of --use-sage-attention:

Faster on Blackwell (RTX 5090 / PRO 6000)
Better detail preservation
No separate SageAttention package needed

⬆️ 2026/09/02 UPDATE ⬆️

The QwenVL node now has a camera_tag dropdown — no more typing [ORBIT] manually in the prompt. Select from 19 camera movements directly in the node UI:

CategoryTagsStatic[STATIC_CAMERA], [LOCKED_OFF]Slow zoom[SLOW_ZOOM_IN], [SLOW_ZOOM_OUT]Fast zoom[FAST_ZOOM_IN], [FAST_ZOOM_OUT]Pan[PAN_LEFT], [PAN_RIGHT]Tilt[TILT_UP], [TILT_DOWN]Dolly[DOLLY_IN], [DOLLY_OUT]Tracking[TRACKING_LEFT], [TRACKING_RIGHT]Crane[CRANE_UP], [CRANE_DOWN]Other[ORBIT], [HANDHELD], [ROLL]

How it works: when you select a tag, it's injected both at the start of the prompt and as a FINAL CAMERA DIRECTIVE at the end — so Qwen 9B actually respects it despite recency bias on long prompts. The tag also gets a short description so Qwen knows exactly what to write.

Subject stays alive: the directive explicitly tells Qwen that the camera tag controls ONLY the camera — the subject must still have natural, lively action (breathing, gestures, expression, body motion) throughout the clip. No more "statue during orbit" problem.

Manual tags still work: if you leave the dropdown on None and type [ORBIT] in your prompt, it's detected and injected automatically as a fallback.

Available on all three QwenVL nodes: AILab_QwenVL, AILab_QwenVL_Advanced, and AILab_QwenVL_PromptEnhancer.

🔄 FL2VA Loop Merged into FL2VA — One Workflow, Bypass Group

The separate MiniMaxH3-Turbo-FL2VA-Loop-Qwen3.5 workflow is removed. The loop trim logic now lives inside the main FL2VA workflow, wrapped in a "Loop Trim" group that can be toggled via the rgthree Fast Groups Bypasser node:

Loop mode (trim active): the ImageFromBatch + ComfyMathExpression nodes trim the frozen tail (~5 frames) for seamless looping
Non-loop mode (trim bypassed): toggle the group off in the Bypasser → VAEDecode passes directly to RIFE/upscale, full frames preserved

No more switching between two workflows — just toggle the group.

🧹 PromptEnhancer Cleanup

Removed the redundant custom_system_prompt input — enhancement_style (presets) + prompt_text (user input) cover all use cases
Removed CUSTOM_ONLY_STYLE ("✍️ Custom Only (no preset)") — no longer needed
Added camera_tag dropdown (same as main QwenVL nodes)

⬆️ 2026/08/31 UPDATE ⬆️

🎲 Wildcards (T2VA Workflow)

The T2VA Turbo workflow includes a WildcardProcessor node that injects randomized prompt fragments from the MadBe's Prompt Engine (__mbe/prmpt/*) wildcard library. Each generation picks a random entry from each wildcard file, so the same seed produces different results across runs unless you pin the seed.

Wildcards Used in the T2VA Workflow

WildcardCategoryWhat it randomizes__mbe/prmpt/vidstyle__Video styleCinematic / anime / vintage film / 3D CG / claymation / watercolor / fantasy etc.__mbe/prmpt/imgcmp/shots__Camera shotClose-up / wide / medium / dolly / crane / orbit / handheld etc.__mbe/prmpt/lctns/rndmlctns__LocationRandom setting (bedroom / beach / studio / alley / forest / rooftop etc.)__mbe/prmpt/light/rndmlight__LightingKey light direction, color temperature, soft/hard, ambient mood__mbe/prmpt/char/favchar__CharacterRandom character archetype (age, body type, hair, ethnicity)__mbe/prmpt/clths/rndmclth__ClothingRandom outfit / garment description__mbe/prmpt/clths/nudty_brsts__NSFWBreast / nudity descriptors (NSFW preset)__mbe/actsolo/brstplng__NSFW actionSolo breast-play action verbs (NSFW preset)

How It Works

The WildcardProcessor node sits before the Qwen3-VL prompt enhancer
At queue time, each __wildcard__ token is replaced with a random line from the corresponding .txt file inside ComfyUI/custom_nodes/ComfyUI-Wildcards/wildcards/mbe/prmpt/
The expanded text is passed to Qwen3-VL, which converts it into the official MiniMax H3 prompt format
Different seed = different wildcard picks — use a fixed seed if you want reproducible results

Customizing Wildcards

Edit existing: open the .txt files under ComfyUI/custom_nodes/ComfyUI-Wildcards/wildcards/mbe/prmpt/ and add/remove lines (one entry per line)
Add your own: create a new .txt file, e.g. mbe/prmpt/mytags.txt, then reference it as __mbe/prmpt/mytags__
Remove a wildcard: delete the __...__ token from the WildcardProcessor text field in the workflow
Disable randomization: replace the __wildcard__ token with a fixed string

Required Custom Node

ComfyUI-TagComplete (includes the WildcardProcessor node and the __mbe/prmpt/* wildcard set) — huchukato/comfy-tagcomplete

The wildcard files ship with the custom node. If a wildcard resolves to empty, the custom node is missing or the wildcard folder is not installed.

🔄 Sampler Change — MiniMaxH3TurboSampler → KSamplerSelect + MiniMaxH3SigmaShift

All 4 Turbo workflows (T2VA, I2VA, FL2VA, R2VA) have been updated to use ComfyUI core nodes instead of the custom MiniMaxH3TurboSampler:

Removed: MiniMaxH3TurboSampler (custom node from Larryvrh/ComfyUI-MiniMax-H3-Turbo)
Added: KSamplerSelect (sampler: euler) + MiniMaxH3SigmaShift (shift_video=12, shift_audio=3) — both ComfyUI core nodes, no custom node required
Scheduler: simple (unchanged)

Why?

On ComfyUI v0.34.2+ with native ModelSamplingAV, the custom MiniMaxH3TurboSampler internally delegates to stock euler anyway — the custom node is redundant
Using core nodes means the same workflow works with both:
- Standard model (minimax_h3_fl2va_pruned_nvfp4_convrot_int8) + Turbo LoRA minimax_h3_turbo_v4_step600_ema
- 10Eros-Max (10Eros_Max_H3_FL2VA-INT8-ConvRot-HQ) + T8 compatibility LoRA minimax_h3_fl2v_turbo_8step_v1.0_10ErosMax_beta1_pruned_compat_v001_T8
Just swap LoadDiffusionModel and LoraLoaderBypassModelOnly — the sampler path stays the same

10Eros-Max (Optional — Experimental)

Model: 10Eros_Max_H3_FL2VA-INT8-ConvRot-HQ.safetensors (~23.5 GB) — DmitryDB/MiniMax-H3-10Eros-Max-Quants
LoRA: minimax_h3_fl2v_turbo_8step_v1.0_10ErosMax_beta1_pruned_compat_v001_T8.safetensors (~1.96 GB) — t8star/minimax_h3_turbo_4step_10ErosMax_test4_pruned_curveproj1025_T8
Sampler: euler + MiniMaxH3SigmaShift (shift 12/3) + simple scheduler — same as standard model
⚠️ NVFP4 degrades quality and audio on the 10Eros fine-tune — use INT8 ConvRot HQ only
⚠️ The T8 LoRA is checkpoint-specific — only works with the exact 10Eros pruned model (SHA-256: f82cc3f723b080e7ae94a7c98f95aa989e387618d0bdc940133dfbd9f432c062)

⬆️ 2026/08/27 UPDATE ⬆️

NVFP4+INT8 ConvRot Hybrid — New Default

Default diffusion models: NVFP4+INT8 ConvRot hybrid (minimax_h3_fl2va_pruned_nvfp4_convrot_int8 / minimax_h3_ref2va_pruned_nvfp4_convrot_int8) from lilcheaty/MiniMax-H3-NVFP4 — NVFP4 on MLP, INT8 ConvRot on attention, BF16 on sensitive layers. Best speed/quality on Blackwell, ~2.5x faster than pure INT8 with identical visual quality
Uncensored text encoder: NVFP4 (qwen3vl_32b_heretic_minimax_h3_nvfp4) from Momoking/Qwen3-VL-32B-Heretic-MiniMax-H3-NVFP4 — fast on Blackwell, no quality impact on text encoding
⚠️ Non-Blackwell GPUs: Use pure INT8 ConvRot models from Comfy-Org/MiniMax-H3 instead. NVFP4 requires sm_120+ (RTX 5090 / PRO 6000).

SOL-ATTN + Spectrum Integration

All 4 Turbo workflows now include SOL-ATTN (Scheduled Sol Attention) for sharper output
All 4 Turbo workflows now include Spectrum adaptive smoothing (offline replay disabled for speed)
Turbo LoRA linked from preset to subgraph in all workflows

Turbo Step Standardization

All Turbo workflows standardized to 8 steps (was 6 for T2VA/R2VA)
Consistent minimax_h3_turbo_v4_step600_ema LoRA across all workflows

TensorRT Batch Size

RIFE and Upscaler TRT expose separate loader and runner batch_size parameters
Verified stable configuration:
- RIFE: loader 1, runner 1
- Upscaler: loader 2, runner 2
The loader compiles the TensorRT engine profile; the runner controls frames sent per infer() call (runner ≤ loader)
RIFE batch values above 1 can build but currently fail during interpolation; keep RIFE at 1/1
Upscaler 2/2 is verified on RTX PRO 6000 Blackwell; batch 4 fails to build with TensorRT 10.15 on sm_120
Changing the loader batch size requires a different engine; delete incompatible cached TRT engines before rebuilding

Upscaler: Auto-detect Scale Factor

Removed the scale dropdown (2x/4x) from the Upscaler runner node — it was redundant and error-prone
The loader now auto-detects the upscale factor from the model name (2x* → 2, 4x* → 4, x2plus → 2, x4plus → 4)
The factor is passed to the runner via the engine object — no more mismatch between model and scale setting
Requires ComfyUI-Upscaler-TensorRT-Auto updated to latest version

⚠️ Requirements — Read First!

GPU & VRAM

🟢 Recommended template configuration — RTX 5090 (32 GB) / RTX PRO 6000 (48 GB) → NVFP4 diffusion + NVFP4 text encoder
🟡 Non-Blackwell alternative — RTX 4090 / 3090 (24 GB) → INT8 ConvRot + offload
🟠 Lower-VRAM alternative — 12–16 GB → INT4 + aggressive offload; slow and not recommended for production
⚠️ NVFP4 requires Blackwell (sm_120+) and does not run on RTX 4090/3090/4080

12 GB GPUs (e.g. RTX 3060 12GB): Technically possible with INT4 models + aggressive offloading, but very slow. You need 32 GB+ system RAM and a fast NVMe SSD. Not recommended for production use.

Model Quantization Options

BF16 (full) — Diffusion ~42 GB + Text encoder ~65 GB = ~110 GB total → Comfy-Org/MiniMax-H3
INT8 (pruned) — Diffusion ~21 GB + Text encoder ~24.5 GB = ~50 GB total → Comfy-Org/MiniMax-H3
INT4 (pruned) — Diffusion ~11 GB + Text encoder ~15 GB = ~24.5 GB total → Merserk/MiniMax-H3-INT4-ConvRot
NVFP4+INT8 ConvRot hybrid (recommended — verified) — Diffusion ~20 GB + uncensored NVFP4 text encoder ~15.7 GB = ~36 GB active model set → lilcheaty/MiniMax-H3-NVFP4 + Momoking/Qwen3-VL-32B-Heretic-MiniMax-H3-NVFP4. Best speed/quality on Blackwell

Software

ComfyUI: v0.31.0+ (required for MiniMax H3 native support)
Python: 3.10+
CUDA: 12.8+ (13.0 recommended)
Storage: allow at least 90 GB for the complete provisioned package (~81 GB of models plus engines, workflows and outputs)

Qwen3-VL Prompt Enhancer

GGUF: Q4_K_S (~4.8 GB) or Q5_K_S (~5.5 GB) for 8B model
HF: Qwen3-VL-8B-Heretic-Stable (~16 GB) or Qwen3-VL-4B (~8 GB)

⚡ MiniMax-H3 Turbo LoRA (Optional — Faster & Sharper)

A distilled 8-step LoRA for MiniMax-H3 that replaces the default ~20-step sampling, with a dedicated ComfyUI node. All Turbo workflows now include SOL-ATTN (Scheduled Sol Attention) and Spectrum adaptive smoothing for sharper, smoother output.

Custom node: Larryvrh/ComfyUI-MiniMax-H3-Turbo
SOL-ATTN node: Saganaki22/ComfyUI-sol-attn
Spectrum node: xmarre/ComfyUI-Spectrum-MiniMax-H3
Recommended LoRA: minimax_h3_turbo_v4_step600_ema.safetensors (~744 MB)
Download: larryvrh/MiniMax-H3-Turbo-Lora
Install: place the .safetensors in ComfyUI/models/loras/
Usage: 8 steps with scheduler simple, SOL-ATTN patch (1.3/0.8/linear/4096), Spectrum (blend_weight=0.5, offline_smoothing_replay=False)

Works with all tasks: T2VA, I2VA, FL2VA and R2VA.

🌟 What is ComfyUI-QwenVL-Mod?

A powerful enhanced vision-language node for ComfyUI that combines Qwen3-VL models with MiniMax H3 video generation workflows. Features multilingual support, visual style detection, native stereo audio, and NSFW capabilities for professional AI content creation.

Think: "Your all-in-one solution for intelligent prompt enhancement and video+audio generation with MiniMax H3!"

🎬 Key Features

🚀 MiniMax H3 Video+Audio Generation

T2VA (Text-to-Video+Audio): Generate video with native stereo audio from text
I2VA (Image-to-Video+Audio): Animate a first-frame image with audio
FL2VA (First-Last-Frame): Generate the transition between two keyframes — Qwen3-VL sees both frames
R2VA (Reference-to-Video): Lock character identity, style, motion, or voice using reference images

🧠 Qwen3-VL Auto-Prompting

Multilingual: Write your prompt in any language — Qwen3-VL translates and converts it
Auto-format: Generates the official MiniMax H3 prompt format (3-field for base, 6-field for R2VA)
Multi-reference: Qwen3-VL sees all connected images via image + image2 inputs
Visual style detection: 12+ artistic styles (photorealistic, cinematic, anime, 3D CG, claymation, vintage film, watercolor, fantasy, etc.)
Smart caching: Performance optimization with Fixed Seed Mode
GGUF backend: Efficient local model inference with quantization support
Qwen3.5 support: Thinking mode disabled via /no_think for fast prompt generation

🔊 Native Stereo Audio

No separate audio node needed — MiniMax H3 generates video and audio jointly in a single forward pass
Voice, sound effects, and music modeled together, not layered on afterward
Describe sounds in your prompt and the model generates them natively

🎨 NSFW Support

Comprehensive content generation without restrictions
9 dedicated NSFW presets (3 base 🎬 + 3 R2VA 🎞️ + 3 FL2VA 🔄) with explicit diegetic soundscape
Natural progression, style adaptation, consistent characters

📦 What's Included — 4 Turbo Workflows

All workflows are pre-wired with MiniMax-H3 Turbo LoRA + MiniMax-H3 Turbo Sampler at 8 steps + SOL-ATTN + Spectrum.

⚡ T2VA Turbo — MiniMaxH3-Turbo-T2VA-Qwen3.5.json — text only — Text-to-video+audio. Simplest workflow.
⚡ I2VA Turbo — MiniMaxH3-Turbo-I2VA-Qwen3.5.json — text + first-frame image (image) — Image-to-video. First-frame animation with audio.
⚡ FL2VA Turbo — MiniMaxH3-Turbo-FL2VA-Qwen3.5.json — text + first-frame (image) + last-frame (image2) — First-Last-Frame to video. Includes TensorRT upscale + RIFE frame interpolation for 48 fps output.
⚡ R2VA Turbo — MiniMaxH3-Turbo-R2VA-Qwen3.5.json — text + reference images (image + image2) — Reference-to-video. Lock identity, style, motion, camera, or voice using up to 9 ref images.

Workflows 3 and 4 include TensorRT upscaling and RIFE frame interpolation for 48 fps high-resolution output.

🖼️ Multi-Reference Input (image2)

The QwenVL-Mod node has two image inputs:

T2VA: no images needed
I2VA: image = first frame
FL2VA: image = first frame, image2 = last frame, frame_count = 1
R2VA: image = primary reference, image2 = additional references (batch, up to 9), frame_count = 1–9

Qwen3-VL sees all connected images as individual images (not as a video sequence), enabling proper multi-reference analysis for FL2VA and R2VA.

🎯 QwenVL-Mod NSFW Presets (9 total)

The workflows include built-in NSFW presets for the Qwen3-VL prompt enhancer:

🎬 Base Presets (T2VA / I2VA)

🎬 MiniMax H3 NSFW (5s) — 5 seconds — 3 fields: integrated_multimodal_description + overall_soundscape + non_diegetic_music
🎬 MiniMax H3 NSFW (10s) — 10 seconds — Same format
🎬 MiniMax H3 NSFW (15s) — 15 seconds — Same format

🔄 FL2VA Presets (First-Last-Frame)

🔄 MiniMax H3 NSFW FL2VA (5s) — 5 seconds — 3 fields, transition-focused (describes the path between frames)
🔄 MiniMax H3 NSFW FL2VA (10s) — 10 seconds — Same format
🔄 MiniMax H3 NSFW FL2VA (15s) — 15 seconds — Same format

🎞️ R2VA Presets (Reference)

🎞️ MiniMax H3 NSFW R2VA (5s) — 5 seconds — 6 fields: subject_definitions + summary + retention_analysis + detailed_description + overall_soundscape + non_diegetic_music
🎞️ MiniMax H3 NSFW R2VA (10s) — 10 seconds — Same format
🎞️ MiniMax H3 NSFW R2VA (15s) — 15 seconds — Same format

What the presets produce

🎬 Base: [Shot 1] with style + initial composition, camera vocabulary, speaker IDs, diegetic soundscape
🔄 FL2VA: Describes the transition path between first and last frames (not the scene — images fix the scene). Favors single continuous shot.
🎞️ R2VA: 6-section format with <Subject N>, <Picture N>, <Video N>, <Audio N> labels, retention markers (fully_preserved, partially_preserved, etc.), task-type summary
All presets: smooth, continuous camera motion (no abrupt or stepped changes), explicit diegetic soundscape, optional non-diegetic music (defaults to N/A)

SFW presets are also available. Edit the preset dropdown in the QwenVL node to switch.

🎮 Usage Examples

Basic Text-to-Video (T2VA)

Load MiniMaxH3-Turbo-T2VA-Qwen3.5.json
Write your prompt in any language
Select preset 🎬 MiniMax H3 NSFW (5s/10s/15s)
Generate video with native audio

Image-to-Video (I2VA)

Load MiniMaxH3-Turbo-I2VA-Qwen3.5.json
Upload your first-frame image to image
Select preset 🎬 MiniMax H3 NSFW (5s/10s/15s)
Write what happens next (in any language)
Generate animated video with audio

First-Last-Frame (FL2VA)

Load MiniMaxH3-Turbo-FL2VA-Qwen3.5.json
Upload first-frame to image, last-frame to image2, set frame_count=1
Select preset 🔄 MiniMax H3 NSFW FL2VA (5s/10s/15s)
Describe the transition between the two frames
Generate the interpolated video at 48 fps with TensorRT upscale + RIFE

Reference-to-Video (R2VA)

Load MiniMaxH3-Turbo-R2VA-Qwen3.5.json
Upload primary reference to image, additional references to image2 (batch), set frame_count to match
Select preset 🎞️ MiniMax H3 NSFW R2VA (5s/10s/15s)
Reference them by tag in your prompt: <Picture 1>, <Picture 2>, etc.
Generate video with locked identity/style

🔧 Technical Specifications

⚡ Performance

Output: 768p, 24 fps (native), up to ~15 seconds
Audio: Native stereo, generated jointly with video
Upscale: TensorRT RealESRGAN x4 (FL2VA + R2VA workflows)
Frame interpolation: RIFE v4.25 → 48 fps (FL2VA + R2VA workflows)
Sage Attention: FP16 accumulation, async offload
Smart caching: Reuse prompts with same inputs, Fixed Seed Mode for text-only caching

🎨 Model Support

Qwen3-VL 4B: 7 GGUF variants (2.38 GB – 4.28 GB)
Qwen3-VL 8B: 7 GGUF variants (4.8 GB – 8.71 GB)
Qwen3.5: 4B / 9B / 27B (uncensored, heretic, unsloth) — thinking mode disabled
HF Models: Josiefed, official, Heretic-Stable variants
Quantization: Q4_K_S, Q5_K_S, FP16, INT8

🌐 Multilingual Capabilities

Input languages: Any language supported
Auto-translation: Automatic translation to optimized English
Style detection: Works with multilingual prompts
Cultural adaptation: Context-aware prompt enhancement

📦 Installation

Quick Install

Download: ComfyUI-QwenVL-Mod (latest version)
Extract to ComfyUI/custom_nodes/ComfyUI-QwenVL-Mod
Install requirements: pip install -r requirements.txt
Restart ComfyUI
Load included workflows from minimax/ folder

Custom Nodes Required

ComfyUI-QwenVL-Mod — All workflows (Qwen3-VL prompt enhancer) — huchukato/ComfyUI-QwenVL-Mod
ComfyUI-MiniMax-H3-Turbo — Turbo workflows (Turbo LoRA + Sampler) — Larryvrh/ComfyUI-MiniMax-H3-Turbo
ComfyUI-sol-attn — Turbo workflows (SOL-ATTN, FusedModulation, ChunkFeedForward) — Saganaki22/ComfyUI-sol-attn
ComfyUI-Spectrum-MiniMax-H3 — Turbo workflows (adaptive smoothing) — xmarre/ComfyUI-Spectrum-MiniMax-H3
ComfyUI-RIFE-TensorRT-Auto — FL2VA, R2VA (frame interpolation) — huchukato/ComfyUI-RIFE-TensorRT-Auto
ComfyUI-Upscaler-TensorRT-Auto — FL2VA, R2VA (upscaling) — huchukato/ComfyUI-Upscaler-TensorRT-Auto
ComfyUI-VideoHelperSuite — FL2VA, R2VA (VHS_VideoCombine) — Kosinkadink/ComfyUI-VideoHelperSuite
ComfyUI-Easy-Use — FL2VA, R2VA (easy showAnything) — yolain/ComfyUI-Easy-Use
comfyui-find-perfect-resolution — All workflows (ResolutionSelector) — ashtar1984/comfyui-find-perfect-resolution

Note: ComfyMathExpression is built into ComfyUI core (v0.24.1+) — no custom node needed.

Models Required

T2VA / I2VA / FL2VA use the NVFP4 FL2VA model (Blackwell GPUs):

models/vae/ → minimax_h3_video_vae_fp16.safetensors (~5 GB)
models/vae/ → minimax_h3_audio_vae_fp32.safetensors (~0.6 GB)
models/diffusion_models/ → minimax_h3_fl2va_pruned_nvfp4_convrot_int8.safetensors (~20 GB) — lilcheaty/MiniMax-H3-NVFP4
models/text_encoders/ → qwen3vl_32b_heretic_minimax_h3_nvfp4.safetensors (~15.7 GB) — Momoking/Qwen3-VL-32B-Heretic-MiniMax-H3-NVFP4

R2VA (ref2va) uses the NVFP4 Ref2VA model:

models/diffusion_models/ → minimax_h3_ref2va_pruned_nvfp4_convrot_int8.safetensors (~20 GB) — lilcheaty/MiniMax-H3-NVFP4

Non-Blackwell GPUs: Use INT8 ConvRot models from Comfy-Org/MiniMax-H3 instead. NVFP4 requires sm_120+ (RTX 5090 / PRO 6000).

10Eros-Max NVFP4 (optional/experimental)

models/diffusion_models/ → 10Eros_Max_H3_FL2VA-INT8-ConvRot-HQ.safetensors (~23.5 GB)
Pair it only with minimax_h3_fl2v_turbo_8step_v1.0_10ErosMax_beta1_pruned_compat_v001_T8.safetensors (~1.96 GB)
Switch both diffusion model and matching LoRA together; do not mix the standard and 10Eros LoRAs
The standard MiniMax H3 NVFP4+INT8 hybrid is the verified default. 10Eros-Max remains experimental; NVFP4 HQ degrades quality and audio on the 10Eros fine-tune, so INT8 ConvRot HQ is used instead

Turbo LoRA (standard — all tasks)

models/loras/ → minimax_h3_turbo_v4_step600_ema.safetensors (~744 MB)
Custom node: Larryvrh/ComfyUI-MiniMax-H3-Turbo

INT4 alternative (for 12-16 GB GPUs): Merserk/MiniMax-H3-INT4-ConvRot

Qwen3-VL Prompt Enhancer

models/LLM/ → Qwen3-VL-8B-Heretic-Stable (GGUF or HF)

TensorRT Engines (FL2VA + R2VA only)

models/upscale_models/ → RealESRGAN_x4 (TensorRT engine)
models/rife/ → rife425_ensemble_False_scale_1_sim (TensorRT engine, ONNX auto-downloaded from HF)

TensorRT engines must be built for your specific GPU. See ComfyUI-RIFE-TensorRT-Auto and ComfyUI-Upscaler-TensorRT-Auto for build instructions.

Download Links

VAE: video_vae_fp16 · audio_vae_fp32
Diffusion (fl2va, NVFP4+INT8 hybrid — Blackwell): minimax_h3_fl2va_pruned_nvfp4_convrot_int8.safetensors
Diffusion (ref2va, NVFP4+INT8 hybrid — Blackwell): minimax_h3_ref2va_pruned_nvfp4_convrot_int8.safetensors
Diffusion (fl2va, INT8 — non-Blackwell): minimax_h3_fl2va_pruned_int8_convrot.safetensors
Diffusion (ref2va, INT8 — non-Blackwell): minimax_h3_ref2va_pruned_int8_convrot.safetensors
Text encoder (uncensored, NVFP4 — Blackwell): qwen3vl_32b_heretic_minimax_h3_nvfp4.safetensors
Text encoder (official, INT8 alternative): qwen3vl_32b_minimax_h3_int8_convrot.safetensors
Turbo LoRA (standard): minimax_h3_turbo_v4_step600_ema.safetensors
10Eros-Max diffusion (optional/experimental, INT8 HQ): 10Eros_Max_H3_FL2VA-INT8-ConvRot-HQ.safetensors
10Eros-compatible Turbo LoRA: minimax_h3_fl2v_turbo_8step_v1.0_10ErosMax_beta1_pruned_compat_v001_T8.safetensors
INT4 models: Merserk/MiniMax-H3-INT4-ConvRot

🎬 MiniMax H3 Prompting Notes

How to Write Your Prompt

Describe the scene naturally. Be clear about the concepts below — Qwen3-VL handles the rest:

🎨 Visual style (put it first): photorealistic, cinematic, anime, 3D CG, claymation, vintage film, watercolor, fantasy
👥 Subjects: number, gender, appearance, clothing, position, expression
🏃 Action / motion: what happens, speed, interaction
🎥 Camera: dolly, pan, zoom, static, handheld, crane, orbit — smooth and continuous (no abrupt changes)
🌍 Environment: setting, lighting, atmosphere, time of day
🔊 Audio (important!): dialogue, breaths, moans, skin contact, ambient sounds, music

🔄 FL2VA: Describe the transition between frames, not the scene (images fix the scene) 🎞️ R2VA: Reference inputs by tag: <Picture 1>, <Picture 2>, <Video 1>, <Audio 1>

Resolution Guidance

MiniMax H3 native canvas: 768 px short edge, long edge capped at 1344 px, multiples of 32.

📱 Portrait: 768×1344 · 896×1152 · 960×1280
⬛ Square: 1024×1024
🖥️ Landscape: 1344×768 · 1152×896 · 1280×960

⚠️ Match the aspect ratio to your input image! Forcing 16:9 on a portrait image will squash it.
⚠️ Avoid direct 1080p. Generate at native resolution, then upscale with TensorRT nodes (FL2VA + R2VA workflows).

Duration

Choose a preset: 5s / 10s / 15s. The Math Expression node snaps the frame count to the model's 17-frame-per-block grid (17k+5 at 24 fps).

🐳 Docker / Cloud Ready

OneClick RunPod Template

Prefer a ready-to-go environment? Use the OneClick - ComfyUI - MiniMax H3 Turbo - Qwen3VL RunPod template:

Docker image: huchukato/comfyui-qwenvl-runpod:cu13-mmh3
Base: huchukato/comfyui-base:cu130
All custom nodes pre-installed
All 4 Turbo workflows auto-downloaded at boot
Models auto-downloaded at first boot (~81 GB including INT8 diffusion, NVFP4 text encoder, 10Eros INT8 HQ and Turbo LoRAs; persistent)
ComfyUI v0.34.2 baked into base image
Sage Attention, FP16 accumulation, async offload
TensorRT upscaling + RIFE interpolation (stable defaults: Upscaler 2/2, RIFE 1/1)
SOL-ATTN + Spectrum for Turbo workflows

Access: ComfyUI :8188 · JupyterLab :8888 · FileBrowser :8080 (user admin / password adminadmin12) · SSH ssh root@pod-ip

📖 README & instructions

ComfyUI Args (pre-configured)

--disable-auto-launch
--fast fp16_accumulation
--use-sage-attention
--reserve-vram 2
--cuda-malloc
--async-offload

🚀 Why Choose ComfyUI-QwenVL-Mod + MiniMax H3?

🎬 For Content Creators

Native audio: Video and audio in one pass — no separate MMAudio needed
Multilingual: Write in any language, Qwen3-VL handles translation
Professional: Official MiniMax H3 prompt format with camera vocabulary and speaker tags
Quality: 768p native, TensorRT upscale to higher resolution

🔥 For NSFW Content

Explicit: Uncensored generation with dedicated NSFW presets
9 presets: 3 base 🎬 + 3 FL2VA 🔄 + 3 R2VA 🎞️ — each tuned for its mode
Detailed: Rich scene descriptions with explicit diegetic soundscape
Natural: Realistic progression, consistent characters
Audio: Native moans, breaths, skin contact, ambient sounds

⚡ For Power Users

Customizable: Easy to modify presets and system prompts
Extendable: Add your own Qwen3-VL models (GGUF or HF)
Integrable: Works with existing ComfyUI setups
Optimized: Sage Attention, FP16, async offload, smart caching
Multi-reference: image2 input for FL2VA and R2VA workflows

🌟 What Makes This Special?

First: Complete MiniMax H3 workflow pack with Qwen3-VL auto-prompting
Native audio: No separate audio node — MiniMax H3 does it all
4 Turbo workflows: T2VA, I2VA, FL2VA, R2VA — covers all MiniMax H3 modes
Multi-reference: Qwen3-VL sees all connected images (not just the first)
TensorRT: Built-in upscaling and frame interpolation
9 NSFW presets: Dedicated presets for each mode with correct prompt structure
Multilingual: Any input language, auto-translated and formatted
Ready: Works out-of-the-box with included workflows

🎯 What's New in v2.6.0

⚡ NVFP4+INT8 ConvRot Hybrid — New Default

✅ Default diffusion models: NVFP4+INT8 ConvRot hybrid (minimax_h3_fl2va_pruned_nvfp4_convrot_int8 / minimax_h3_ref2va_pruned_nvfp4_convrot_int8)
✅ ~2.5x faster than pure INT8 ConvRot with identical visual quality on Blackwell
✅ NVFP4 on MLP, INT8 ConvRot on attention, BF16 on sensitive layers (by rockerBOO/lilcheaty)
✅ NVFP4 requires Blackwell GPUs (RTX 5090 / PRO 6000, sm_120+)
✅ Pure INT8 ConvRot models still available for non-Blackwell GPUs

🔧 SOL-ATTN + Spectrum Integration

✅ All 4 Turbo workflows now include SOL-ATTN (Scheduled Sol Attention) for sharper output
✅ All 4 Turbo workflows now include Spectrum adaptive smoothing (offline replay disabled for speed)
✅ Turbo LoRA linked from preset to subgraph in all workflows

⚡ Turbo Step Standardization

✅ All Turbo workflows standardized to 8 steps (was 6 for T2VA/R2VA)
✅ Consistent minimax_h3_turbo_v4_step600_ema LoRA across all workflows

📦 TensorRT Batch Size

✅ Stable defaults: RIFE loader/runner 1/1, Upscaler loader/runner 2/2
✅ Upscaler 2/2 verified on RTX PRO 6000 Blackwell
⚠️ RIFE batch >1 currently fails during interpolation even when the engine builds; keep it at 1/1
⚠️ Upscaler batch 4 fails to build with TensorRT 10.15 on Blackwell sm_120

🧹 Cleanup

✅ Removed was-node-suite from MiniMax Dockerfile (not used by MiniMax workflows)
✅ Removed ComfyUI-Frame-Interpolation from MiniMax (uses RIFE TensorRT instead)
✅ Removed KJNodes from provisioning (baked into RunPod base image)
✅ ComfyMathExpression is built into ComfyUI core — no custom node needed

🧠 Qwen3.5 Thinking Fix

✅ /no_think prefix for Qwen3.5 models (enable_thinking deprecated in recent llama.cpp)
✅ Broadened architecture detection (qwen35, qwen35moe, qwen35_vl)
✅ Works across both HF and GGUF nodes

📦 Workflow Organization

✅ Moved workflows to minimax/ folder
✅ Renamed FLF to FL2VA (clearer naming)
✅ Added Civitai documentation

📋 Credits

MiniMax H3 — MiniMax · Comfy-Org/MiniMax-H3
ComfyUI — comfyanonymous/ComfyUI
QwenVL-Mod — huchukato/ComfyUI-QwenVL-Mod
Qwen3-VL — Qwen Team / Alibaba
INT4 models — Merserk/MiniMax-H3-INT4-ConvRot
TensorRT RIFE / Upscaler — huchukato
SOL-ATTN — Saganaki22
Spectrum — xmarre
VideoHelperSuite — Kosinkadink
Easy-Use — yolain
find-perfect-resolution — ashtar1984

📄 License

Workflows are released under the same license as the underlying models and custom nodes. See each repository for details.

MiniMax H3 model weights: Comfy-Org/MiniMax-H3 — MiniMax H3 Community License.

Built with ❤️ for the ComfyUI community

Files with SHA256: 9a6dc0c1f8a20ce2f51b58f0756937dc7bf2f78fd0643b772d6cce9afa2d0557