OpenAI-compatible API
Expose local model chat completions through a familiar HTTP API for existing clients and tools.
Rust AI runtime
Werk1112 – Local-first AI inference router in Rust
Werk1112 is a local-first AI runtime and inference router for llama.cpp, vLLM, ONNX Runtime, Candle and MLX with an OpenAI-compatible API.
Local model routing with one API surface
Applications target Werk1112 while Werk1112 chooses the local or companion runtime for the model format, requested backend, and host capabilities.
Runtime planner
Route GGUF, safetensors, ONNX, and MLX-style model directories to the appropriate backend.
Local-first workflow
Manage imports, pulls, estimates, CLI chat, and serving without making the app depend on one engine.
Project topics
Werk1112 focuses on Rust-based local AI infrastructure and inference routing for common LLM runtimes.
- Rust
- AI
- LLM
- local AI
- inference router
- OpenAI-compatible API
- llama.cpp
- vLLM
- ONNX Runtime
- Candle
- MLX
- local-first AI runtime
Source code
Werk1112 is developed in the philipbodenbach/werk1112 GitHub repository by Philip Bodenbach.