Doesnt load/run the openVINO model droans/qwen3.6-27B-int4-asym-ov on NPU on V3.1.0

## Describe the bug
NPU chat keeps on loading for a while but doesnt actually load/run the droans/qwen3.6-27B-int4-asym-ov model.
i have NOT try this model on other component that is CPU/GPU/iGPU.

## To Reproduce

i open NPU chat tab and provided the droans/qwen3.6-27B-int4-asym-ov repo to download the model from huggingface. it downloads successfully. i chose Vision capability and NPU support. typed 32768 context size.

## Expected behavior

Load and run/infer the droans/qwen3.6-27B-int4-asym-ov as it is exported to openVINO type

## Screenshots

[https://drive.google.com/file/d/1XgwPnVi7LBRRCko8wa3YL9Hva8cdVid4/view?usp=sharing](Screen recording of the Ai Playground App)

## Environment (please complete the following information):

- OS: Windows11
- GPU: Laptop RTX 5080
- CPU: ultra 9 275HX
- Version: v3.1.0

Start in the NPU chat tab and reproduce the failure on Windows 11 with v3.1.0 using the Hugging Face model droans/qwen3.6-27B-int4-asym-ov, Vision capability, NPU support, and a 32768 context size. Check the model-loading and inference path; done means the model loads and runs inference on the NPU.

Written by the indexing model from the issue text.