After first generation, Cuda error: Invalid Argument and cannot generate again until restart

[Volcanicus]

Custom Node Testing

• I have tried disabling custom nodes and the issue persists (see how to disable custom nodes if you need help)

Your question

I checked one of the threads with the same Cuda error but neither the --disable-pinned-memory nor updating fixed the issue.

I came back after a break and decided to go to the manager and update all.
I then encountered a malloc.py error of sorts that prevented the program from starting up.
I checked around and decided to fix it by loading up an older .py file from a backup. That didn't work.
I ran the update_comfyui.bat and it seemingly fixed the startup issue.
I went back and updated again and it finally generated.
However, when I try to queue something else, I get "got prompt" and the queue shows a new item but nothing else happens.

This happens on a basic workflow, custom, WAN 2.1, WAN 2.2 and whatever else. Basically the first gen works but anything subsequent doesn't go through.
I also tried to have multiple gens per queue but after the first gen it will simply not go through.

Here are the console logs:

D:\AI\ComfyUI_windows_portable>.\python_embeded\python.exe -s ComfyUI\main.py --windows-standalone-build --fast
[START] Security scan
[DONE] Security scan

ComfyUI-Manager: installing dependencies done.

** ComfyUI startup time: 2026-02-18 23:28:23.079
** Platform: Windows
** Python version: 3.12.9 (tags/v3.12.9:fdb8142, Feb 4 2025, 15:27:58) [MSC v.1942 64 bit (AMD64)]
** Python executable: D:\AI\ComfyUI_windows_portable\python_embeded\python.exe
** ComfyUI Path: D:\AI\ComfyUI_windows_portable\ComfyUI
** ComfyUI Base Folder Path: D:\AI\ComfyUI_windows_portable\ComfyUI
** User directory: D:\AI\ComfyUI_windows_portable\ComfyUI\user
** ComfyUI-Manager config path: D:\AI\ComfyUI_windows_portable\ComfyUI\user__manager\config.ini
** Log path: D:\AI\ComfyUI_windows_portable\ComfyUI\user\comfyui.log

Prestartup times for custom nodes:
0.0 seconds: D:\AI\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-easy-use
2.2 seconds: D:\AI\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-manager

WARNING: You need pytorch with cu130 or higher to use optimized CUDA operations.
Found comfy_kitchen backend eager: {'available': True, 'disabled': False, 'unavailable_reason': None, 'capabilities': ['apply_rope', 'apply_rope1', 'dequantize_nvfp4', 'dequantize_per_tensor_fp8', 'quantize_nvfp4', 'quantize_per_tensor_fp8', 'scaled_mm_nvfp4']}
Found comfy_kitchen backend triton: {'available': True, 'disabled': True, 'unavailable_reason': None, 'capabilities': ['apply_rope', 'apply_rope1', 'dequantize_nvfp4', 'dequantize_per_tensor_fp8', 'quantize_nvfp4', 'quantize_per_tensor_fp8']}
Found comfy_kitchen backend cuda: {'available': True, 'disabled': True, 'unavailable_reason': None, 'capabilities': ['apply_rope', 'apply_rope1', 'dequantize_nvfp4', 'dequantize_per_tensor_fp8', 'quantize_nvfp4', 'quantize_per_tensor_fp8']}
Checkpoint files will always be loaded safely.
Total VRAM 16376 MB, total RAM 32655 MB
pytorch version: 2.8.0.dev20250317+cu128
Enabled fp16 accumulation.
Set vram state to: NORMAL_VRAM
Device: cuda:0 NVIDIA GeForce RTX 4070 Ti SUPER : native
Using async weight offloading with 2 streams
Enabled pinned memory 14694.0
Using pytorch attention
aimdo: src/control.c:51:INFO:comfy-aimdo inited for GPU: NVIDIA GeForce RTX 4070 Ti SUPER (VRAM: 16375 MB)
DynamicVRAM support detected and enabled
Python version: 3.12.9 (tags/v3.12.9:fdb8142, Feb 4 2025, 15:27:58) [MSC v.1942 64 bit (AMD64)]
ComfyUI version: 0.14.1
ComfyUI frontend version: 1.38.14
[Prompt Server] web root: D:\AI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\comfyui_frontend_package\static
[ComfyUI-Easy-Use] server: v1.3.6 Loaded
[ComfyUI-Easy-Use] web root: D:\AI\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-easy-use\web_version/v2 Loaded
ComfyUI-GGUF: Partial torch compile only, consider updating pytorch

Loading: ComfyUI-Impact-Pack (V8.28.2)

[Impact Pack] Wildcard total size (0.00 MB) is within cache limit (50.00 MB). Using full cache mode.
[Impact Pack] Wildcards loading done.

Loading: ComfyUI-Inspire-Pack (V1.23)

---

[ComfyUI-Manager] NOTICE: Legacy backup exists

• Your old Manager data was backed up to:
  D:\AI\ComfyUI_windows_portable\ComfyUI\user__manager.legacy-manager-backup
• Please verify and remove it when no longer needed.

---

Loading: ComfyUI-Manager (V3.39.2)

[ComfyUI-Manager] network_mode: public
[ComfyUI-Manager] ComfyUI per-queue preview override detected (PR #11261). Manager's preview method feature is disabled. Use ComfyUI's --preview-method CLI option or 'Settings > Execution > Live preview method'.

ComfyUI Revision: 4747 [185c61d] *DETACHED | Released on '2026-02-18'

[MultiGPU Core Patching] Patching mm.soft_empty_cache for Comprehensive Memory Management (VRAM + CPU + Store Pruning)
[MultiGPU Core Patching] Patching mm.get_torch_device, mm.text_encoder_device, mm.unet_offload_device
[MultiGPU DEBUG] Initial current_device: cuda:0
[MultiGPU DEBUG] Initial current_text_encoder_device: cuda:0
[MultiGPU DEBUG] Initial current_unet_offload_device: cpu
[MultiGPU] Initiating custom_node Registration. . .

custom_node Found Nodes

ComfyUI-LTXVideo N 0
ComfyUI-Florence2 N 0
ComfyUI_bitsandbytes_NF4 N 0
x-flux-comfyui N 0
ComfyUI-MMAudio N 0
ComfyUI-GGUF Y 18
PuLID_ComfyUI N 0
ComfyUI-WanVideoWrapper Y 20

[MultiGPU] Registration complete. Final mappings: CheckpointLoaderAdvancedMultiGPU, CheckpointLoaderAdvancedDisTorch2MultiGPU, UNetLoaderLP, UNETLoaderMultiGPU, VAELoaderMultiGPU, CLIPLoaderMultiGPU, DualCLIPLoaderMultiGPU, TripleCLIPLoaderMultiGPU, QuadrupleCLIPLoaderMultiGPU, CLIPVisionLoaderMultiGPU, CheckpointLoaderSimpleMultiGPU, ControlNetLoaderMultiGPU, DiffusersLoaderMultiGPU, DiffControlNetLoaderMultiGPU, UNETLoaderDisTorch2MultiGPU, VAELoaderDisTorch2MultiGPU, CLIPLoaderDisTorch2MultiGPU, DualCLIPLoaderDisTorch2MultiGPU, TripleCLIPLoaderDisTorch2MultiGPU, QuadrupleCLIPLoaderDisTorch2MultiGPU, CLIPVisionLoaderDisTorch2MultiGPU, CheckpointLoaderSimpleDisTorch2MultiGPU, ControlNetLoaderDisTorch2MultiGPU, DiffusersLoaderDisTorch2MultiGPU, DiffControlNetLoaderDisTorch2MultiGPU, UnetLoaderGGUFDisTorchMultiGPU, UnetLoaderGGUFAdvancedDisTorchMultiGPU, CLIPLoaderGGUFDisTorchMultiGPU, DualCLIPLoaderGGUFDisTorchMultiGPU, TripleCLIPLoaderGGUFDisTorchMultiGPU, QuadrupleCLIPLoaderGGUFDisTorchMultiGPU, UnetLoaderGGUFDisTorch2MultiGPU, UnetLoaderGGUFAdvancedDisTorch2MultiGPU, CLIPLoaderGGUFDisTorch2MultiGPU, DualCLIPLoaderGGUFDisTorch2MultiGPU, TripleCLIPLoaderGGUFDisTorch2MultiGPU, QuadrupleCLIPLoaderGGUFDisTorch2MultiGPU, UnetLoaderGGUFMultiGPU, UnetLoaderGGUFAdvancedMultiGPU, CLIPLoaderGGUFMultiGPU, DualCLIPLoaderGGUFMultiGPU, TripleCLIPLoaderGGUFMultiGPU, QuadrupleCLIPLoaderGGUFMultiGPU, LoadWanVideoT5TextEncoderMultiGPU, WanVideoTextEncodeMultiGPU, WanVideoTextEncodeCachedMultiGPU, WanVideoTextEncodeSingleMultiGPU, WanVideoVAELoaderMultiGPU, WanVideoTinyVAELoaderMultiGPU, WanVideoBlockSwapMultiGPU, WanVideoImageToVideoEncodeMultiGPU, WanVideoDecodeMultiGPU, WanVideoModelLoaderMultiGPU, WanVideoSamplerMultiGPU, WanVideoVACEEncodeMultiGPU, WanVideoEncodeMultiGPU, LoadWanVideoClipTextEncoderMultiGPU, WanVideoClipVisionEncodeMultiGPU, WanVideoControlnetLoaderMultiGPU, FantasyTalkingModelLoaderMultiGPU, Wav2VecModelLoaderMultiGPU, WanVideoUni3C_ControlnetLoaderMultiGPU, DownloadAndLoadWav2VecModelMultiGPU
[ComfyUI-Manager] default cache updated: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/alter-list.json
[ComfyUI-Manager] default cache updated: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/model-list.json
[ComfyUI-Manager] default cache updated: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/extension-node-map.json
[ComfyUI-Manager] default cache updated: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/github-stats.json
[ComfyUI-Manager] default cache updated: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/custom-node-list.json
[D:\AI\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui_controlnet_aux] | INFO -> Using ckpts path: D:\AI\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui_controlnet_aux\ckpts
[D:\AI\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui_controlnet_aux] | INFO -> Using symlinks: False
[D:\AI\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui_controlnet_aux] | INFO -> Using ort providers: ['CUDAExecutionProvider', 'DirectMLExecutionProvider', 'OpenVINOExecutionProvider', 'ROCMExecutionProvider', 'MIGraphXExecutionProvider', 'CPUExecutionProvider', 'CoreMLExecutionProvider']
DWPose: Onnxruntime with acceleration providers detected
WAS Node Suite: BlenderNeko's Advanced CLIP Text Encode found, attempting to enable CLIPTextEncode support.
WAS Node Suite: CLIPTextEncode (BlenderNeko Advanced + NSP) node enabled under WAS Suite/Conditioning menu.
WAS Node Suite: OpenCV Python FFMPEG support is enabled
WAS Node Suite Warning: ffmpeg_bin_path is not set in D:\AI\ComfyUI_windows_portable\ComfyUI\custom_nodes\was-ns\was_suite_config.json config file. Will attempt to use system ffmpeg binaries if available.
WAS Node Suite: Finished. Loaded 221 nodes successfully.

    "The artist's world is limitless. It can be found anywhere, far from where he lives or a few feet away. It is always on his doorstep." - Paul Strand

Import times for custom nodes:
0.0 seconds: D:\AI\ComfyUI_windows_portable\ComfyUI\custom_nodes\websocket_image_save.py
0.0 seconds: D:\AI\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI_ADV_CLIP_emb
0.0 seconds: D:\AI\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Adaptive-Guidance
0.0 seconds: D:\AI\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-unload-model
0.0 seconds: D:\AI\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-GGUF
0.0 seconds: D:\AI\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI_essentials
0.0 seconds: D:\AI\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Frame-Interpolation
0.0 seconds: D:\AI\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-KJNodes
0.0 seconds: D:\AI\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-multigpu
0.0 seconds: D:\AI\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-inspire-pack
0.0 seconds: D:\AI\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-VideoHelperSuite
0.1 seconds: D:\AI\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-impact-pack
0.2 seconds: D:\AI\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui_controlnet_aux
0.5 seconds: D:\AI\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-manager
0.7 seconds: D:\AI\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-WanVideoWrapper
1.0 seconds: D:\AI\ComfyUI_windows_portable\ComfyUI\custom_nodes\was-ns
2.4 seconds: D:\AI\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-easy-use

Context impl SQLiteImpl.
Will assume non-transactional DDL.
Assets scan(roots=['models']) completed in 0.055s (created=0, skipped_existing=427, orphans_pruned=0, total_seen=428)
Starting server

To see the GUI go to: http://127.0.0.1:8188
[DEPRECATION WARNING] Detected import of deprecated legacy API: /scripts/ui.js. This is likely caused by a custom node extension using outdated APIs. Please update your extensions or contact the extension author for an updated version.
[DEPRECATION WARNING] Detected import of deprecated legacy API: /extensions/core/clipspace.js. This is likely caused by a custom node extension using outdated APIs. Please update your extensions or contact the extension author for an updated version.
[DEPRECATION WARNING] Detected import of deprecated legacy API: /extensions/core/groupNode.js. This is likely caused by a custom node extension using outdated APIs. Please update your extensions or contact the extension author for an updated version.
[DEPRECATION WARNING] Detected import of deprecated legacy API: /extensions/core/widgetInputs.js. This is likely caused by a custom node extension using outdated APIs. Please update your extensions or contact the extension author for an updated version.
[Inspire Pack] IPAdapterPlus is not installed.
FETCH ComfyRegistry Data: 5/126
[DEPRECATION WARNING] Detected import of deprecated legacy API: /scripts/ui/components/buttonGroup.js. This is likely caused by a custom node extension using outdated APIs. Please update your extensions or contact the extension author for an updated version.
[DEPRECATION WARNING] Detected import of deprecated legacy API: /scripts/ui/components/button.js. This is likely caused by a custom node extension using outdated APIs. Please update your extensions or contact the extension author for an updated version.
FETCH ComfyRegistry Data: 10/126
FETCH ComfyRegistry Data: 15/126
FETCH ComfyRegistry Data: 20/126
FETCH ComfyRegistry Data: 25/126
FETCH ComfyRegistry Data: 30/126
FETCH ComfyRegistry Data: 35/126
FETCH ComfyRegistry Data: 40/126
FETCH ComfyRegistry Data: 45/126
FETCH ComfyRegistry Data: 50/126
FETCH ComfyRegistry Data: 55/126
FETCH ComfyRegistry Data: 60/126
FETCH ComfyRegistry Data: 65/126
FETCH ComfyRegistry Data: 70/126
FETCH ComfyRegistry Data: 75/126
FETCH ComfyRegistry Data: 80/126
FETCH ComfyRegistry Data: 85/126
FETCH ComfyRegistry Data: 90/126
FETCH ComfyRegistry Data: 95/126
FETCH ComfyRegistry Data: 100/126
FETCH ComfyRegistry Data: 105/126
FETCH ComfyRegistry Data: 110/126
FETCH ComfyRegistry Data: 115/126
FETCH ComfyRegistry Data: 120/126
FETCH ComfyRegistry Data: 125/126
FETCH ComfyRegistry Data [DONE]
[ComfyUI-Manager] default cache updated: https://api.comfy.org/nodes
FETCH DATA from: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/custom-node-list.json [DONE]
[ComfyUI-Manager] All startup tasks have been completed.
got prompt
model weight dtype torch.float16, manual cast: None
model_type EPS
Using pytorch attention in VAE
Using pytorch attention in VAE
VAE load device: cuda:0, offload device: cpu, dtype: torch.bfloat16
[MultiGPU Core Patching] text_encoder_device_patched returning device: cuda:0 (current_text_encoder_device=cuda:0)
Requested to load SDXLClipModel
Model SDXLClipModel prepared for dynamic VRAM loading. 1560MB Staged. 0 patches attached.
CLIP/text encoder model load device: cuda:0, offload device: cpu, current: cuda:0, dtype: torch.float16
model_path is D:\AI\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui_controlnet_aux\ckpts\lllyasviel\Annotators\body_pose_model.pth
model_path is D:\AI\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui_controlnet_aux\ckpts\lllyasviel\Annotators\hand_pose_model.pth
model_path is D:\AI\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui_controlnet_aux\ckpts\lllyasviel\Annotators\facenet.pth
Requested to load SDXLClipModel
Model SDXLClipModel prepared for dynamic VRAM loading. 1560MB Staged. 264 patches attached.
Model SDXLClipModel prepared for dynamic VRAM loading. 1560MB Staged. 264 patches attached.
Model SDXLClipModel prepared for dynamic VRAM loading. 1560MB Staged. 264 patches attached.
Requested to load SDXL
Requested to load ControlNet
Model SDXL prepared for dynamic VRAM loading. 4897MB Staged. 722 patches attached.
Model ControlNet prepared for dynamic VRAM loading. 2386MB Staged. 0 patches attached.
100%|██████████████████████████████████████████████████████████████████████████████████| 35/35 [00:07<00:00, 4.62it/s]
Requested to load AutoencoderKL
Model AutoencoderKL prepared for dynamic VRAM loading. 159MB Staged. 0 patches attached.
Requested to load SDXL
Model SDXL prepared for dynamic VRAM loading. 4897MB Staged. 0 patches attached.
100%|██████████████████████████████████████████████████████████████████████████████████| 20/20 [00:08<00:00, 2.25it/s]
0 models unloaded.
Model AutoencoderKL prepared for dynamic VRAM loading. 159MB Staged. 0 patches attached.
Prompt executed in 36.18 seconds
Exception in thread Thread-10 (prompt_worker):
Traceback (most recent call last):
File "threading.py", line 1075, in _bootstrap_inner
File "threading.py", line 1012, in run
File "D:\AI\ComfyUI_windows_portable\ComfyUI\main.py", line 302, in prompt_worker
comfy.model_management.soft_empty_cache()
File "D:\AI\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-multigpu\device_utils.py", line 258, in soft_empty_cache_distorch2_patched
original_soft_empty_cache(force)
File "D:\AI\ComfyUI_windows_portable\ComfyUI\comfy\model_management.py", line 1712, in soft_empty_cache
torch.cuda.empty_cache()
File "D:\AI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\cuda\memory.py", line 222, in empty_cache
torch._C._cuda_emptyCache()
RuntimeError: CUDA error: invalid argument
CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1
Compile with TORCH_USE_CUDA_DSA to enable device-side assertions.

got prompt

========================================================================

Logs

Other

No response

[GlamoramaAttack]

"After first generation, Cuda error: Invalid Argument and cannot generate again until restart"

Happens to me too, with Qwen Image Edit (2509/2511), Flux2 (dev+klein dist/base) and the main problem is, if I accidentally use the wrong CLIP and the message appears, it also appears after plugging the correct one in. I have to restart ComfyUI because whatever wrong CLIP it tried to start, it doesn't unload without a restart. I also had to restart the computer in a few cases, because generated images looked distorted, it seemed something was still left in the VRAM. If I accidentally use the wrong main model or the wrong vae and exchange them the CUDA error doesn't happen and there is no need for a restart.

pytorch version: 2.10.0+cu130
Device: cuda:0 NVIDIA GeForce RTX 3050 : cudaMallocAsync
Using async weight offloading with 2 streams
Using pytorch attention
Python version: 3.13.12
ComfyUI version: 0.13.0
ComfyUI frontend version: 1.38.13

Main difference to user Volcanicus (except for different versions), I didn't have a malloc.py error, don't use Multi-GPU and don't have this warning:

WARNING: You need pytorch with cu130 or higher to use optimized CUDA operations.

[Volcanicus]

I ended up running the full updater "update_comfyui_and_python_dependencies.bat" and it seems to have fixed the issue.