# walrus-model
LLM provider implementations for Walrus.
Supports DeepSeek, OpenAI-compatible, Claude, and local inference (via
mistral.rs). Includes `ProviderManager` for multi-provider routing and
`ProviderConfig` for configuration.
## Features
- `local` (default) — Local model inference via mistral.rs
- `cuda` — NVIDIA CUDA GPU acceleration
- `metal` — Apple Metal GPU acceleration
## License
GPL-3.0