multi-model-example the result is incorrect


**Which Component**
multi-model-example, /video

**Describe the bug**
Qwen vl max latest recognition example video, but interpreted as an image.

**Expected behavior**
It should be interpreted as a video

**Screenshots**
![video20241222_112515](https://github.com/user-attachments/assets/355c9838-e6b8-4dbe-bf0a-198fa38c0230)

**Additional context**
Win11  、Java17 、 Version 1.0.0-M3.2