After completing the installation of vllm, simply downgrade the transformers version.
The commands you need are:

pip install -U vllm
pip install transformers-v4.55.0-GLM-4.5V-preview

Check dependencies with:

pip list
pip show vllm transformers

It is also recommended to clear the pip cache:

pip cache purge

When launching this model with vllm, the required command is:

vllm serve \
    QuantTrio/GLM-4.5V-AWQ \
    --served-model-name GLM-4.5V-AWQ \
    --enable-expert-parallel \
    --tensor-parallel-size 4   # replace 4 with the actual number of GPUs

I follow the guide to install env:

Patched vLLM (see: https://github.com/vllm-project/vllm/pull/22716)

git clone -b glm-45 https://github.com/zRzRzRzRzRzRzR/vllm.git
cd vllm
VLLM_USE_PRECOMPILED=1 pip install .

Install preview build of Transformers with GLM-4.5V support

pip install transformers-v4.55.0-GLM-4.5V-preview

but still got the error

INFO 08-23 17:18:09 [init.py:241] Automatically detected platform cuda.
Traceback (most recent call last):
File "/mnt/workspace/zichen.shx/infer/infer_v6_cmd_glm.py", line 25, in
from vllm import LLM, SamplingParams
File "", line 1075, in _handle_fromlist
File "/opt/conda/envs/glmfp4/lib/python3.10/site-packages/vllm/init.py", line 64, in getattr
module = import_module(module_name, package)
File "/opt/conda/envs/glmfp4/lib/python3.10/importlib/init.py", line 126, in import_module
return _bootstrap._gcd_import(name[level:], package, level)
File "/opt/conda/envs/glmfp4/lib/python3.10/site-packages/vllm/entrypoints/llm.py", line 24, in
from vllm.engine.llm_engine import LLMEngine
File "/opt/conda/envs/glmfp4/lib/python3.10/site-packages/vllm/engine/llm_engine.py", line 28, in
from vllm.engine.output_processor.util import create_output_by_sequence_group
File "/opt/conda/envs/glmfp4/lib/python3.10/site-packages/vllm/engine/output_processor/util.py", line 8, in
from vllm.model_executor.layers.sampler import SamplerOutput
File "/opt/conda/envs/glmfp4/lib/python3.10/site-packages/vllm/model_executor/layers/sampler.py", line 16, in
from vllm.model_executor.layers.utils import apply_penalties
File "/opt/conda/envs/glmfp4/lib/python3.10/site-packages/vllm/model_executor/layers/utils.py", line 8, in
from vllm import _custom_ops as ops
File "/opt/conda/envs/glmfp4/lib/python3.10/site-packages/vllm/_custom_ops.py", line 472, in
def _marlin_qqq_gemm_fake(a: torch.Tensor, b_q_weight: torch.Tensor,
File "/opt/conda/envs/glmfp4/lib/python3.10/site-packages/torch/library.py", line 1023, in register
use_lib._register_fake(op_name, func, _stacklevel=stacklevel + 1)
File "/opt/conda/envs/glmfp4/lib/python3.10/site-packages/torch/library.py", line 214, in _register_fake
handle = entry.fake_impl.register(func_to_register, source)
File "/opt/conda/envs/glmfp4/lib/python3.10/site-packages/torch/_library/fake_impl.py", line 31, in register
if torch._C._dispatch_has_kernel_for_dispatch_key(self.qualname, "Meta"):
RuntimeError: operator _C::marlin_qqq_gemm does not exist

Thank you for your feedback. This error occurs a lot in the recent nightly versions.
At the moment of this post, we can just install the official vllm:

pip install vllm==0.10.1.1
pip install transformers-v4.55.0-GLM-4.5V-preview

I have updated the readme file accordingly.

thank you for your answer, it does work!