vLLM
The High-Throughput and Memory-Efficient inference and serving engine for LLMs
Local inference
Open vLLM in AirStackWhat vLLM’s own pages say
| What it says it does | The High-Throughput and Memory-Efficient inference and serving engine for LLMs vllm.ai |
|---|---|
| Platforms | NVIDIA CUDA GPU, AMD GPUs, NVIDIA GPUs, x86/ARM/PowerPC CPUs vllm.ai docs.vllm.ai |
| Integrations | Hugging Face models docs.vllm.ai |
| Open source | Yes docs.vllm.ai |
| Models supported | DeepSeek V3.2, DeepSeek V4, DeepSeek V4.1 Flash, Gemma, Llama, Qwen vllm.ai docs.vllm.ai |
Each value is a quote AirStack found on the product’s own pages, read 2026-11-04. A missing field means AirStack has not found it on its pages.