All plugins
(AI) — Extend
Local Models
Run open-weight language models on your own GPU and hand them to Crown Agent, with no data leaving the machine.
$ crown plugin add local-models- Author
- CrownOS Team
- Version
- 1.4.0
- Category
- AI
- Tags
- llm, ollama, offline, gpu
Local Models pulls, quantises and serves open-weight models through a single system daemon. Crown Agent and any plugin that asks for a model get it over a local socket, so prompts and files never leave the machine.
What it does
- Detects your GPU and picks the right backend: CUDA, ROCm or Vulkan
- Downloads models on demand and shares one copy between apps
- Unloads idle models after a timeout to free VRAM
- Exposes an OpenAI-compatible endpoint on
localhostfor your own tools
Permissions
| Scope | Why |
|---|---|
| GPU | Runs inference on your graphics card |
| Network (opt-in) | Downloads model weights |