Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add znlgis/opengis-skills --skill vllmgit clone --depth 1 https://github.com/znlgis/opengis-skillsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/znlgis/opengis-skills/vllm)<a href="https://agentmods.dev/skills/znlgis/opengis-skills/vllm"><img src="https://agentmods.dev/badge/skills/znlgis/opengis-skills/vllm/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/znlgis/opengis-skills/vllm"><img src="https://agentmods.dev/badge/skills/znlgis/opengis-skills/vllm.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00121 | $0.04346 |
| Opus 5.5 | $0.00048 | $0.01738 |
| Sonnet 5.5 | $0.00024 | $0.00869 |
| Haiku 4.5 | $0.00012 | $0.00435 |
Grade A, and why
vllm scanned grade A with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured today.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
curl http://localhost:8000/v1/models How it starts
The opening of the file, as written. The whole thing — 334 lines — stays where its author put it; the contents beside it link to each section on GitHub.
项目地址: https://github.com/vllm-project/vllm
官方文档: https://docs.vllm.ai/
PyPI:
vllm(0.32.0,截至 2026-10)许可证: Apache-2.0
概述
vLLM 是目前最主流的开源 LLM 推理与服务平台,把"高吞吐、低延迟、OpenAI 兼容"三件事一次做掉。
- PagedAttention:KV cache 按页管理,显存碎片少、并发高
- Continuous batching / chunked prefill / prefix caching:连续批处理、分块预填充、前缀复用
- 量化:FP8、MXFP8/MXFP4、NVFP4、INT8、INT4、GPTQ/AWQ、GGUF、compressed-tensors、ModelOpt、TorchAO 等
- 并行:tensor / pipeline / data / expert / context 并行,支持多机
- OpenAI 兼容 API:
/v1/models、/v1/completions、/v1/chat/completions,另支持 Anthropic Messages API 与 gRPC - 模型面广:200+ Hugging Face 架构(Llama / Qwen / Gemma / Mixtral / DeepSeek-V3 / Qwen-MoE / GPT-OSS / Mamba / 多模态 / embedding / reward)
- 硬件:NVIDIA、AMD、Intel GPU,x86/ARM/PowerPC CPU,以及 TPU / Gaudi / Ascend 等插件
环境准备
前置条件
| 项 | 要求 |
|---|---|
| 操作系统 | Linux(macOS 需走 vLLM-Metal;Windows 建议 WSL2 + Linux 容器) |
| Python | 3.11 – 3.14 |
| GPU | NVIDIA 需匹配的 CUDA 驱动;AMD 走 ROCm wheel;Intel 走 XPU |
| 磁盘 | 模型权重 + 编译产物,几十 GB 起步 |
方式一:uv(官方推荐)
uv venv --python 3.12 --seed
source .venv/bin/activate
uv pip install vllm --torch-backend=auto
vllm download-kernels
--torch-backend=auto 会按驱动版本自动选 PyTorch 源;也可写死 --torch-backend=cu126。vllm download-kernels 预编译 kernel,避免首次启动在 Hopper 及以上卡上编译几分钟。
方式二:pip / conda
# pip
python -m pip install vllm
# conda
conda create -n vllm python=3.12 -y
conda activate vllm
pip install --upgrade uv
uv pip install vllm --torch-backend=auto
方式三:不改环境直接试
uv run --with vllm vllm --help
其他硬件
# AMD ROCm(支持 Python 3.12、ROCm 7.0、glibc >= 2.35)
uv pip install vllm --extra-index-url https://wheels.vllm.ai/rocm/
# Google TPU
uv pip install vllm-tpu
Apple Silicon 用 vLLM-Metal;Ascend NPU 用 vllm-ascend。
验证
vllm --version
vllm serve --help | head -30
离线批量推理
from vllm import LLM, SamplingParams
prompts = [
"Hello, my name is",
"The president of the United States is",
"The capital of France is",
"The future of AI is",
]
sampling_params = SamplingParams(temperature=0.8, top_p=0.95)
llm = LLM(model="facebook/opt-125m")
outputs = llm.generate(prompts, sampling_params)
for output in outputs:
print(f"Prompt: {output.prompt!r}")
print(f"Generated: {output.outputs[0].text!r}")
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- today First seen · 334 lines · 121 tokens per session scan A 994628e3ee1d
vllm is a skill published in the GitHub repository znlgis/opengis-skills (75 stars, last pushed today), licensed MIT. It adds 121 tokens to every session and 4,346 once invoked, about $0.0005 per session on Opus 5.5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-10-28.
Other skills, from other repositories
groq-inference
Ultra-fast LLM inference on custom LPU hardware. OpenAI-compatible API at api.groq.com. Lowest latency in the industry (500-1000+ tok/s). Supports chat completions, vision, audio (Whisper STT + TTS), tool calling, JSON mode, and streaming. Free tier available. Inference only — no training.
plugin-system
Generic plugin system for Python applications. Auto-discovery, validation, fault tolerance. Zero dependencies (Python stdlib only).
hqq-quantization
Half-Quadratic Quantization for LLMs without calibration data. Use when quantizing models to 4/3/2-bit precision without needing calibration datasets, for fast quantization workflows, or when deploying with vLLM or HuggingFace Transformers.
instructor
Extract structured data from LLM responses with Pydantic validation, retry failed extractions automatically, parse complex JSON with type safety, and stream partial results with Instructor - battle-tested structured output library.
pytorch-patterns
PyTorch deep learning patterns and best practices for building robust, efficient, and reproducible training pipelines, model architectures, and data loading.
django-patterns
Django architecture patterns, REST API design with DRF, ORM best practices, caching, signals, middleware, and production-grade Django apps.