Feature Request: Update internal Ollama binary to support new Qwen architectures

[JPetovello]

The current image throws a "needs a newer version of Ollama" error when trying to pull newer models like qwen3:4b

[shansen008]

Would love this too, but I think this has been abandoned.

I have a Pro B50, its such a great card with potential, but if you want to run newer models youre stuck with Vulkan....ewww. Intels specialized hardware is nice, but hardly anyone codes for it.