A desktop GUI application for managing an OpenVINO Model Server (OVMS) instance on Windows. Built with Python 3.12 and CustomTkinter.
Download the latest installer from Releases and run it. The app installs to:
%LOCALAPPDATA%\Programs\OVMS Manager\
All runtime data (models, workspace, logs, config) is stored under:
%LOCALAPPDATA%\OVMS Manager\
models\ downloaded model files
ovms\ OVMS binary (auto-installed via Setup tab)
workspace\ OVMS config.json and graph.pbtxt
env\ Python venv (auto-created via Setup tab)
logs\ server and proxy log files
config.json app settings
On first launch the app asks whether to auto-start the OVMS stack on every open. If any components are missing, the Setup tab guides you through installing them.
pip install -r requirements.txt
python main.pyor double-click run.bat (auto-detects Python from %USERPROFILE%\openvino-env or PATH).
- Status cards — live green/red indicators for OVMS, proxy, and active model.
- Start Stack / Stop Stack — single button to start or stop both processes.
- Log viewer — last 25 lines of the OVMS log, auto-refreshed every 2 seconds.
- Curated list of OpenVINO-optimised coding and reasoning models.
- Per-row Download button — downloads from HuggingFace Hub with a live progress bar.
- Activate button — rewrites
graph.pbtxtandconfig.json, then restarts OVMS.
| Display Name | HuggingFace Repo | Size | Notes |
|---|---|---|---|
| Qwen2.5-Coder-7B | OpenVINO/Qwen2.5-Coder-7B-Instruct-int4-ov | ~4 GB | Best coding |
| Qwen3-8B | OpenVINO/Qwen3-8B-int4-ov | ~4 GB | General |
| Qwen3-8B (NPU) | OpenVINO/Qwen3-8B-int4-cw-ov | ~4 GB | NPU optimised |
| DeepSeek-R1-7B | OpenVINO/DeepSeek-R1-Distill-Qwen-7B-int4-cw-ov | ~4 GB | Reasoning |
| DeepSeek-R1-1.5B | OpenVINO/DeepSeek-R1-Distill-Qwen-1.5B-int4-cw-ov | ~1 GB | Fast, lightweight |
| Phi-3.5-mini | OpenVINO/Phi-3.5-mini-instruct-int4-cw-ov | ~2 GB | Efficient |
main.py Entry point, logging setup
app/
gui.py CustomTkinter window, tabs, status cards, model rows
server.py OVMS + proxy subprocess management, health polling
models.py Model catalog, HuggingFace download, config writing
config.py JSON-backed settings, all paths under %LOCALAPPDATA%
installer.py Component detection and auto-installation
log_viewer.py Auto-refreshing log tail widget
workspace\graph.pbtxtis rewritten with the selected model's local path.workspace\config.jsonis rewritten to reference that graph.- OVMS is restarted to pick up the new configuration.
OVMS serves on REST port 8000; the proxy (OpenAI-compatible) serves on port 8001.