MiniMax H3 NVFP4 model loading issue on RTX5090D

### Custom Node Testing

- [x] I have tried disabling custom nodes and the issue persists (see [how to disable custom nodes](https://docs.comfy.org/troubleshooting/custom-node-issues#step-1%3A-test-with-all-custom-nodes-disabled) if you need help)

### Expected Behavior

When I launched the portable comfyui to create video with MiniMax-H3 template, and change the model to minimax_h3_fl2va_pruned_nvfp4.safetensors, which could run smoothly and succeed.

### Actual Behavior

When I launched the portable comfyui to create video with MiniMax-H3 template, and change the model to minimax_h3_fl2va_pruned_nvfp4.safetensors, which could run smoothly and succeed. But the log showed on the model **Model Initializing ...**

### Steps to Reproduce

1. Launch the portable comfyui;
2. Load the minimax-h3-text-to-video tempalte.
3. Enter the prompts.
4. Change the unet_name to **minimax_h3_fl2va_pruned_nvfp4.safetensors**
5. Run.

### Debug Logs

```powershell
[INFO] setup plugin alembic.autogenerate.schemas
[INFO] setup plugin alembic.autogenerate.tables
[INFO] setup plugin alembic.autogenerate.types
[INFO] setup plugin alembic.autogenerate.constraints
[INFO] setup plugin alembic.autogenerate.defaults
[INFO] setup plugin alembic.autogenerate.comments
[INFO] Found comfy_kitchen backend triton: {'available': False, 'disabled': True, 'unavailable_reason': "ImportError: No module named 'triton'", 'capabilities': []}
[INFO] Found comfy_kitchen backend hip: {'available': False, 'disabled': False, 'unavailable_reason': 'PyTorch ROCm/HIP runtime not available', 'capabilities': []}
[INFO] Found comfy_kitchen backend eager: {'available': True, 'disabled': False, 'unavailable_reason': None, 'capabilities': ['adaln', 'apply_rope', 'apply_rope1', 'apply_rope1_', 'apply_rope_', 'apply_rope_split_half', 'apply_rope_split_half1', 'apply_rope_split_half1_', 'apply_rope_split_half_', 'convrot_w4a4_linear', 'dequantize_convrot_w4a4_weight', 'dequantize_int8_convrot_weight', 'dequantize_int8_convrot_weight_dtype', 'dequantize_int8_embedding', 'dequantize_int8_simple', 'dequantize_int8_simple_dtype', 'dequantize_mxfp8', 'dequantize_nvfp4', 'dequantize_per_tensor_fp8', 'dequantize_w4a8_int8_weight', 'gemv_awq_w4a16', 'int8_linear', 'na3d', 'prepare_int4_weight_for_int8_linear', 'quantize_and_rotate_rowwise', 'quantize_convrot_w4a4_weight', 'quantize_int8_convrot_weight', 'quantize_int8_rowwise', 'quantize_int8_tensorwise', 'quantize_mxfp8', 'quantize_nvfp4', 'quantize_per_tensor_fp8', 'quantize_svdquant_w4a4', 'quantize_w4a8_int8_weight', 'rms_adaln', 'rms_rope', 'rms_rope1', 'rms_rope1_', 'rms_rope_', 'rms_rope_split_half', 'rms_rope_split_half1', 'rms_rope_split_half1_', 'rms_rope_split_half_', 'rotate_int8_convrot_weight', 'scaled_mm_mxfp8', 'scaled_mm_nvfp4', 'scaled_mm_svdquant_w4a4', 'stochastic_rounding_fp8', 'w4a8_int8_linear']}
[INFO] Found comfy_kitchen backend cuda: {'available': True, 'disabled': False, 'unavailable_reason': None, 'capabilities': ['adaln', 'apply_rope', 'apply_rope1', 'apply_rope1_', 'apply_rope_', 'apply_rope_split_half', 'apply_rope_split_half1', 'apply_rope_split_half1_', 'apply_rope_split_half_', 'convrot_w4a4_linear', 'dequantize_convrot_w4a4_weight', 'dequantize_int8_convrot_weight', 'dequantize_int8_convrot_weight_dtype', 'dequantize_int8_simple', 'dequantize_int8_simple_dtype', 'dequantize_nvfp4', 'dequantize_per_tensor_fp8', 'dequantize_w4a8_int8_weight', 'gemv_awq_w4a16', 'int8_linear', 'na3d', 'prepare_int4_weight_for_int8_linear', 'quantize_and_rotate_rowwise', 'quantize_convrot_w4a4_weight', 'quantize_int8_convrot_weight', 'quantize_int8_rowwise', 'quantize_int8_tensorwise', 'quantize_mxfp8', 'quantize_nvfp4', 'quantize_per_tensor_fp8', 'quantize_svdquant_w4a4', 'quantize_w4a8_int8_weight', 'rms_adaln', 'rms_rope', 'rms_rope1', 'rms_rope1_', 'rms_rope_', 'rms_rope_split_half', 'rms_rope_split_half1', 'rms_rope_split_half1_', 'rms_rope_split_half_', 'rotate_int8_convrot_weight', 'scaled_mm_nvfp4', 'scaled_mm_svdquant_w4a4', 'stochastic_rounding_fp8', 'w4a8_int8_linear']}
[INFO] Checkpoint files will always be loaded safely.
[INFO] Total VRAM 32607 MB, total RAM 65149 MB
[INFO] pytorch version: 2.13.0+cu130
[INFO] Set vram state to: NORMAL_VRAM
[INFO] Device: cuda:0 NVIDIA GeForce RTX 5090 D : cudaMallocAsync
[INFO] Using async weight offloading with 2 streams
[INFO] Enabled pinned memory 26059.0
[INFO] Using pytorch attention
[INFO] aimdo: src-win/cuda-detour.c:38:INFO:aimdo_setup_hooks: installing 6 hooks
[INFO] aimdo: src/control.c:262:INFO:comfy-aimdo NVML pressure enabled
[INFO] aimdo: src-win/shmem-detect.c:80:INFO:comfy-aimdo WDDM adapter match: NVIDIA GeForce RTX 5090 D runtime_luid=00000000:2918b458 dxgi_luid=00000000:2918b458
[INFO] aimdo: src/control.c:277:INFO:comfy-aimdo inited for GPU: NVIDIA GeForce RTX 5090 D (VRAM: 32606 MB)
[INFO] DynamicVRAM support detected and enabled
[INFO] Python version: 3.13.14 (tags/v3.13.14:fd17997, Jun 10 2026, 13:03:48) [MSC v.1944 64 bit (AMD64)]
[INFO] ComfyUI version: 0.33.1
[INFO] comfy-aimdo version: 0.4.13
[INFO] comfy-kitchen version: 0.2.31
[INFO] comfyui-frontend-package version: 1.48.7
[INFO] comfyui-workflow-templates version: 0.11.41
[INFO] comfyui-embedded-docs version: 0.5.9
[INFO] comfy-kitchen version: 0.2.31
[INFO] comfy-aimdo version: 0.4.13
[INFO] [Prompt Server] web root: D:\AI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\comfyui_frontend_package\static
[INFO] Asset seeder disabled
[INFO] No OpenGL_accelerate module loaded: No module named 'OpenGL_accelerate'
[INFO]
Import times for custom nodes:
[INFO] 0.0 seconds: D:\AI\ComfyUI_windows_portable\ComfyUI\custom_nodes\websocket_image_save.py
[INFO] 0.0 seconds: D:\AI\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-MiniMax-H3-Guide
[INFO] 0.1 seconds: D:\AI\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-minimax-h3-audio-T8
[INFO]
[INFO] Context impl SQLiteImpl.
[INFO] Will assume non-transactional DDL.
[INFO] Using RAM pressure cache.
[INFO] Starting server

[INFO] To see the GUI go to: http://127.0.0.1:8188
[INFO] got prompt
[INFO] VAE load device: cuda:0, offload device: cpu, dtype: torch.float32
[INFO] VAE load device: cuda:0, offload device: cpu, dtype: torch.float16
[INFO] Found quantization metadata version 1
[INFO] Using MixedPrecisionOps for text encoder
[INFO] CLIP/text encoder model load device: cuda:0, offload device: cpu, current: cpu, dtype: torch.float16
[INFO] Requested to load MiniMaxH3TEModel_
[INFO] Model MiniMaxH3TEModel_ prepared for dynamic VRAM loading. 14956MB Staged. 0 patches attached. Force pre-loaded 410 weights: 4572 KB.
[INFO] Interrupting prompt c5a7ff90-e9f8-4a96-bc41-db82ead7ca7a
[INFO] Processing interrupted
[INFO] Prompt executed in 1.88 seconds
[INFO] got prompt
[INFO] Model MiniMaxH3TEModel_ prepared for dynamic VRAM loading. 14956MB Staged. 0 patches attached. Force pre-loaded 410 weights: 4572 KB.
[INFO] Found quantization metadata version 1
[INFO] Detected mixed precision quantization
[INFO] Using mixed precision operations
[INFO] Native ops: float8_e4m3fn, convrot_w4a4, float8_e5m2, mxfp8, int8_tensorwise, nvfp4, asym_w4a8_int8
[INFO] model weight dtype torch.bfloat16, manual cast: torch.bfloat16
[INFO] model_type FLOW_AV
[INFO] Requested to load MiniMaxH3
[INFO] 0 models unloaded.
[INFO] Model MiniMaxH3 prepared for dynamic VRAM loading. 11944MB Staged. 0 patches attached. Force pre-loaded 210 weights: 1175 KB.
0%| | 0/20 [00:00

Start by reproducing the portable ComfyUI workflow with the MiniMax-H3 text-to-video template, the minimax_h3_fl2va_pruned_nvfp4.safetensors unet_name, and the RTX 5090 D environment described in the log. Check the model-loading stage after the workflow reaches 0/20 and “Model Initializing ...”. Done means the workflow completes video generation rather than remaining at initialization.

Written by the indexing model from the issue text.