[Bug] ComfyUI Flux Error: string pointer is null

[healthyfat]

Checklist

• 1. I have searched for related issues and FAQs (https://nunchaku.tech/docs/nunchaku/faq/faq.html) but was unable to find a solution.
• 2. The issue persists in the latest version.
• 3. Please note that without environment information and a minimal reproducible example, it will be difficult for us to reproduce and address the issue, which may delay our response.
• 4. If your report is a question rather than a bug, please submit it as a discussion at https://github.com/mit-han-lab/nunchaku/discussions/new/choose. Otherwise, this issue will be closed.
• 5. If this is related to ComfyUI, please report it at https://github.com/mit-han-lab/ComfyUI-nunchaku/issues.
• 6. I will do my best to describe the issue in English.

Describe the Bug

After updating to Nunchaku v1.2.0 ComfyUI displays following error.

Previous working version was v1.0.2.

ComfyUI Error Report

Error Details

• Node ID: 45
• Node Type: NunchakuFluxDiTLoader
• Exception Type: RuntimeError
• Exception Message: string pointer is null

got prompt
Failed to validate prompt for output 118:

• (prompt):
  • Required input is missing: source
• Display Any (rgthree) 118:
  • Required input is missing: source
    Output will be ignored
    Failed to validate prompt for output 117:
• (prompt):
  • Required input is missing: text
• ShowText|pysssss 117:
  • Required input is missing: text
    Output will be ignored
    WARNING: PlaySound.IS_CHANGED() missing 1 required positional argument: 'self'
    Using pytorch attention in VAE
    Using pytorch attention in VAE
    VAE load device: cuda:0, offload device: cpu, dtype: torch.float32
    GPU 0 (NVIDIA GeForce RTX 2060) Memory: 6143.6875 MiB
    VRAM < 14GiB, enabling CPU offload
    [2026-01-12 15:18:33.914] [info] Initializing QuantizedFluxModel on device 0
    [2026-01-12 15:18:33.915] [info] Use FP16 model
    [2026-01-12 15:18:33.915] [info] Layer offloading enabled
    !!! Exception during processing !!! string pointer is null
    Traceback (most recent call last):
    File "D:\AI\UI\ComfyUI_windows_portable\ComfyUI\execution.py", line 518, in execute
    output_data, output_ui, has_subgraph, has_pending_tasks = await get_output_data(prompt_id, unique_id, obj, input_data_all, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb, v3_data=v3_data)
    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    File "D:\AI\UI\ComfyUI_windows_portable\ComfyUI\execution.py", line 329, in get_output_data
    return_values = await _async_map_node_over_list(prompt_id, unique_id, obj, input_data_all, obj.FUNCTION, allow_interrupt=True, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb, v3_data=v3_data)
    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    File "D:\AI\UI\ComfyUI_windows_portable\ComfyUI\execution.py", line 303, in _async_map_node_over_list
    await process_inputs(input_dict, i)
    File "D:\AI\UI\ComfyUI_windows_portable\ComfyUI\execution.py", line 291, in process_inputs
    result = f(**inputs)
    ^^^^^^^^^^^
    File "D:\AI\UI\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-nunchaku\nodes\models\flux.py", line 283, in load_model
    self.transformer, self.metadata = NunchakuFluxTransformer2dModel.from_pretrained(
    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    File "D:\AI\UI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\huggingface_hub\utils_validators.py", line 114, in _inner_fn
    return fn(*args, **kwargs)
    ^^^^^^^^^^^^^^^^^^^
    File "D:\AI\UI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\nunchaku\models\transformers\transformer_flux.py", line 605, in from_pretrained
    m = load_quantized_module(
    ^^^^^^^^^^^^^^^^^^^^^^
    File "D:\AI\UI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\nunchaku\models\transformers\transformer_flux.py", line 454, in load_quantized_module
    m.init(use_fp4, offload, bf16, 0 if device.index is None else device.index)
    RuntimeError: string pointer is null

Environment

Name: nunchaku
Version: 1.2.0+torch2.9

System Information

• ComfyUI Version: 0.8.0
• Arguments: ComfyUI\main.py --windows-standalone-build
• OS: win32
• Python Version: 3.12.7 (tags/v3.12.7:0b05ead, Oct 1 2024, 03:06:41) [MSC v.1941 64 bit (AMD64)]
• Embedded Python: true
• PyTorch Version: 2.9.0+cu128

Devices

• Name: cuda:0 NVIDIA GeForce RTX 2060 : cudaMallocAsync
  • Type: cuda
  • VRAM Total: 6442123264
  • VRAM Free: 5355077632
  • Torch VRAM Total: 0
  • Torch VRAM Free: 0
  • Driver version: 576.57

Reproduction Steps

svdq-int4_r32-flux.1-dev.safetensors

[WhiteCrowLX]

Just switch back to the 1.0.2 wheel. I don't know why, but I also have this problem. However, if I can't use the 1.2.0 wheel, I can't use the Zimage model. I wonder if any expert can solve this? Thank you.

[healthyfat]

That is what I did. The latest working version for me is v1.0.2. Was hoping for that Turing (20~ Series) GPU compatibility for Z Image Turbo in v1.2.0 though.

[pm-cz]

The general suggestion is to upgrade to latest NVIDIA drivers 591.xx from their website on Windows as a workaround, there is something which does not work (compile?) in the standard ones supplied by Microsoft, see ComfyUI-nunchaku #741

[github-actions]

This issue has been automatically closed due to 30-day inactivity. Please feel free to reopen it with /reopen if needed.