[Doc] Add Qwen3-Omni-30B-A3B-Thinking Tutorials

[Meihan-chen]

What this PR does / why we need it?

Add Qwen3-Omni-30B-A3B-Thinking Tutorials

Does this PR introduce any user-facing change?

No

How was this patch tested?

• vLLM version: v0.11.0
• vLLM main: vllm-project/vllm@83f478b

[github-actions]

👋 Hi! Thank you for contributing to the vLLM Ascend project. The following points will speed up your PR merge:‌‌

• A PR should do only one thing, smaller PRs enable faster reviews.
• Every PR should include unit tests and end-to-end tests ‌to ensure it works and is not broken by other future PRs.
• Write the commit message by fulfilling the PR description to help reviewer and future developers understand.

If CI fails, you can run linting and testing checks locally according Contributing and Testing.

[gemini-code-assist]

Code Review

This PR adds a new tutorial for running Qwen3-Omni-30B-A3B-Thinking on multiple NPUs. The documentation is comprehensive, but I've found a few critical issues in the provided code snippets that would prevent users from successfully running the examples. Specifically, there's an incorrect package name in a pip install command and an incorrect model path in the offline inference script. There is also a typo in the filename of the new document. Please address these issues to ensure the tutorial is accurate and easy to follow.

[gemini-code-assist]

The package name for the Hugging Face transformers library is transformers (plural), not transformer. The current command will fail because the package transformer does not exist on PyPI.

Suggested change

	# pip install transformer -U
	# pip install transformers -U

[gemini-code-assist]

The MODEL_PATH is set to an absolute path "/Qwen/Qwen3-Omni-30B-A3B-Thinking". This is inconsistent with other tutorials and the online inference command in this same file, which use the model identifier directly. Using an absolute path might cause model loading to fail if the model is not present at that exact location in the container. It's better to use the model identifier and let vLLM handle the download and caching, especially since VLLM_USE_MODELSCOPE=True is set.

Suggested change

	MODEL_PATH = "/Qwen/Qwen3-Omni-30B-A3B-Thinking"
	MODEL_PATH = "Qwen/Qwen3-Omni-30B-A3B-Thinking"

[gemini-code-assist]

There's a typo in the filename. It should be multi_npu_qwen3_omni_30B_A3B_Thinking.md instead of multi_npu_qwen3_omni_30B_A3B_Thingking.md. Please correct the filename for consistency and clarity.

[wangxiyuan]

please following the change 5f08e07

We're working on tutorial refactor now.