Skip to content

vLLM #2047

Description

@MickLesk

Name of the Script

vLLM

Script Type

CT (LXC Container)

Does this script support arm64?

arm64 not supported

📋 Script Details

vLLM is a high-throughput, memory-efficient inference and serving engine for LLMs, exposing an OpenAI-compatible HTTP API (/v1) on port 8000 so existing OpenAI clients can serve local models. Installed via uv into a venv at /opt/vllm; the target model is set via VLLM_MODEL in /opt/vllm/vllm.env and downloaded from HuggingFace on first start. NVIDIA GPU passthrough is recommended for usable performance (vllm was added to the GPU passthrough app list; CPU-only inference is impractically slow).

Metadata

Metadata

Assignees

No one assigned

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions