Provide a mock instance to test vLLM without CUDA or any GPUs.

• Github repository: https://github.com/vkehfdl1/vllm-mock/

• vllm.LLM.generate mock
• vllm.LLM.chat mock

It is highly recommended to use the mock instance with pytest-mock.

from vllm_mock import LLM
from vllm import SamplingParams

def test_vllm(mocker):
    mock_class = mocker.patch("vllm.LLM")
    mock_class.return_value = LLM(model="mock-model")

    llm = mock_class()
    sampling_params = SamplingParams(temperature=0.8, top_p=0.9, logprobs=1)
    response = llm.generate("Hello, world!", sampling_params=sampling_params)
    assert isinstance(response[0].outputs[0].text, str)

    chat_response = llm.chat([
		{"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "Hello, world!"}
    ], sampling_params=sampling_params)
    assert isinstance(chat_response[0].outputs[0].text, str)

pip install vllm-mock pytest-mock

or in a UV environment

uv add --dev vllm-mock pytest-mock

• Mock vLLM API server
• Mock Reasoning model features
• Mock quantization features
• Mock LoRA features
• vLM models mock
• vllm.LLM.beam_search mock
• vllm.LLM.embed mock
• vllm.LLM.classify mock
• vllm.LLM.encode mock
• vllm.LLM.reward mock

First, clone a repository

git clone https://github.com/NomaDamas/vllm-mock.git
cd vllm-mock

Then, install the environment and the pre-commit hooks with

make install

This will also generate your uv.lock file

Initially, the CI/CD pipeline might be failing due to formatting issues. To resolve those run:

uv run pre-commit run -a

You can create any issue or PR to support this project. Thank you!

• Jeffrey is a creator of this repo. Made this because he desperately needed it for his research.
• NomaDamas is an AI open-source Hacker House in Seoul, Korea.

Repository initiated with fpgmaas/cookiecutter-uv.