Comfy crash OOM

### 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

generate

### Actual Behavior

crash

### Steps to Reproduce

Just generate (I use VLM model with Diffusing model)

### Debug Logs

```powershell
0: 640x512 1 face, 43.2ms
Speed: 10.1ms preprocess, 43.2ms inference, 22.0ms postprocess per image at shape (1, 3, 640, 512)
Detailer: segment skip [determined upscale factor=0.8890318870544434]
Prompt executed in 238.19 seconds
got prompt
We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
Loading checkpoint shards: 100%|█████████████████████████████████████████████████████████| 4/4 [00:09<00:00, 2.42s/it]
Some parameters are on the meta device because they were offloaded to the cpu.
deal image
deal messages [{'role': 'user', 'content': [{'type': 'image', 'image': }, {'type': 'text', 'text': 'Describe the image in detail.'}]}]
Requested to load QwenImageTEModel_
loaded partially; 6968.65 MB usable, 6968.65 MB loaded, 7807.76 MB offloaded, lowvram patches: 0

C:\GenBot\ComfyUI_windows_portable>pause
```

### Other

_No response_

The report names no source file, test, or specific entry point. Start by reproducing generation with the stated VLM and diffusion models while reviewing the memory and offloading messages in the debug log. Done would require identifying the OOM trigger and confirming that generation no longer crashes under the reported setup.

Written by the indexing model from the issue text.