Metadata-Version: 2.3
Name: k-openvino
Version: 0.1.0
Summary: OpenVINO GenAI OpenAI-compatible server + CLI (sovereign local LLM)
Author: KAMDEM Ivann
Author-email: KAMDEM Ivann <kapoivha@gmail.com>
License: MIT
Requires-Dist: openvino-genai==2026.3.1
Requires-Dist: optimum-intel==2.1.0
Requires-Dist: transformers
Requires-Dist: fastapi
Requires-Dist: uvicorn
Requires-Dist: huggingface-hub
Requires-Dist: httpx
Requires-Dist: typer
Requires-Dist: rich
Requires-Dist: python-dotenv
Requires-Python: >=3.12
Description-Content-Type: text/markdown

# openvino — Sovereign Local LLM (OpenVINO GenAI)

OpenAI-compatible server + Ollama-like CLI for Qwen3 and other text models exported to OpenVINO IR. Intel Arc iGPU (Vulkan) ready, 100% `uv`, `typer` + `rich`, no `sh` wrappers.

## Stack

- Python 3.12, `uv`, `openvino-genai==2026.3.1`, `optimum-intel==2.1.0`, `transformers`, `fastapi` + `uvicorn`, `huggingface_hub`, `httpx`, `typer`, `rich`, `python-dotenv`
- `src/openvino/config.py` — plain frozen dataclass (no YAML)
- `CONTRACT.md` — source of truth

## Install

```bash
# From project root
uv sync --python 3.12
make uv-link   # or: uv tool install --editable . --force
openvino --help
```

Or directly:

```bash
uv tool install --editable . --force
```

## Usage

```bash
openvino ls                          # list local IR models
openvino ps                          # show loaded model
openvino pull qwen3:1.7b            # HF Qwen/Qwen3-1.7B → IR
openvino pull Qwen/Qwen3-4B         # HF id direct
openvino rm qwen3:1.7b
openvino search qwen3
openvino show qwen3:1.7b
openvino run qwen3:1.7b "Bonjour"

# Server (systemd)
systemctl --user enable --now openvino.service
curl -s http://127.0.0.1:11436/health | jq
curl -s http://127.0.0.1:11436/v1/models | jq

# Env
OPENVINO_HOME=~/.local/share/openvino  # models root
OPENVINO_URL=http://127.0.0.1:11436
OPENVINO_HOST=127.0.0.1:11436   # or :11436 (Ollama-style)
```

## Development

```bash
make check   # ruff + pyright + pytest
make help
```
