Linux command
inference-snaps 命令
文本
复制后可按需替换文件名、目录或参数。
常用示例
Example
inference-snaps chat
Example
inference-snaps status
Example
sudo inference-snaps use-engine cuda
Example
sudo inference-snaps show-machine
说明
inference-snaps (and the associated model snaps such as `deepseek-r1`, `gemma3`, `gemma4`, etc.) provide a simple way to run powerful open-weight LLMs locally on Ubuntu without writing any code or managing Python environments. The snaps bundle the model weights, an inference engine (CPU, CUDA, ROCm, etc.), and a small chat server. Once started, you can talk to the model from the terminal or connect other tools to the local HTTP API. Supported models (as of 2026) include DeepSeek R1, Google Gemma 3/4, Nemotron, Qwen-VL and others. Each model snap installs its own command that behaves like `inference-snaps`.
FAQ
What is the inference-snaps command used for?
inference-snaps (and the associated model snaps such as `deepseek-r1`, `gemma3`, `gemma4`, etc.) provide a simple way to run powerful open-weight LLMs locally on Ubuntu without writing any code or managing Python environments. The snaps bundle the model weights, an inference engine (CPU, CUDA, ROCm, etc.), and a small chat server. Once started, you can talk to the model from the terminal or connect other tools to the local HTTP API. Supported models (as of 2026) include DeepSeek R1, Google Gemma 3/4, Nemotron, Qwen-VL and others. Each model snap installs its own command that behaves like `inference-snaps`.
How do I run a basic inference-snaps example?
Run `inference-snaps chat` in a terminal, then adjust file names, paths, flags, or remote targets for your system.
Where can I find more inference-snaps examples?
This page includes 4 examples for inference-snaps, plus related commands for nearby Linux tasks.