v2.0.3
ModelPark CLI
Expose local apps and APIs to the internet in seconds
Quick Install (macOS / Linux)
curl -fsSL https://raw.githubusercontent.com/model-park/cli/master/install.sh | shWindows users: see the installation section below for PowerShell instructions.
Installation
Install the ModelPark CLI to expose your local ML applications to the internet.
🐧 Linux Installation
1
Run the installer script (auto-detects architecture)
curl -fsSL https://raw.githubusercontent.com/model-park/cli/master/install.sh | sh2
Verify installation
modelpark versionOr download manually:
Quick Start
Get your local application online in three steps.
1
Authenticate
Log in to ModelPark using your account credentials or API token.
modelpark login2
Run your app
Start your app and create a tunnel in one step. The port is auto-detected for known frameworks.
modelpark run --name myapp -- streamlit run app.py3
Access your app
Your app is live. Open the remote URL to access it from anywhere.
https://myapp.modelpark.appREST API Quick Start
1
Authenticate
Log in to ModelPark using your account credentials or API token.
modelpark login2
Serve your API
Expose your REST API with API key authentication enabled.
modelpark serve --name myapi --port 8000 --type api3
Call your API
Use your API key to make authenticated requests from anywhere.
curl https://myapi-proxy.modelpark.app/health -H "X-ModelPark-Key: YOUR_KEY"Or run
modelpark with no arguments for an interactive guided experience, or add --type api for REST APIs.Commands
Complete reference for all CLI commands. Use modelpark --help to see available commands.
modelpark runStart a local app and tunnel it
Start your application and create a tunnel in one step. The port is auto-detected for known frameworks — just pass your start command after --. Runs in the background by default; use --session to keep it in the foreground.
Options
| Flag | Short | Default | Description |
|---|---|---|---|
| --name*required | -n | - | App name / subdomain |
| --port | -p | auto-detect | Local port (0 = auto-detect from framework) |
| --access | - | private | Access type (public, private, protected) |
| --type | - | auto-detect | App type: api, web (auto-detected from framework) |
| --token | -t | - | Authentication token (optional if logged in) |
| --session | -s | false | Run in foreground (session mode) |
| --timeout | - | 60s | Max wait for port ready |
| --server | - | wss://tunnel.modelpark.app/ws | Tunnel server WebSocket URL |
| --api-server | - | https://modelpark.app | API server URL |
Supported Frameworks (auto-detect)
| Framework | Default Port | Type |
|---|---|---|
| Streamlit | 8501 | Web |
| Gradio | 7860 | Web |
| Uvicorn | 8000 | API |
| FastAPI | 8000 | API |
| Flask | 5000 | API |
| Gunicorn | 8000 | API |
| Chainlit | 8000 | Web |
| Panel | 5006 | Web |
| Voila | 8866 | Web |
| Jupyter | 8888 | Web |
| MLflow | 5000 | Web |
| Ollama | 11434 | API |
Examples
# Streamlit — port auto-detected as 8501 (runs in background)modelpark run --name myapp -- streamlit run app.py# Keep in foreground (session mode)modelpark run -n myapp --session -- streamlit run app.py# Gradio with public accessmodelpark run -n demo --access public -- python gradio_app.py# Custom port for an unknown frameworkmodelpark run -n myapp -p 9000 -- python server.py# REST API with auto-detected typemodelpark run --name myapi -- uvicorn main:app --port 8000# Explicitly set API typemodelpark run --name myapi --type api -- python server.pyOutput (background mode)
Detected framework: streamlit (port 8501)
Registering app 'myapp'...
App 'myapp' registered (private)
Started 'myapp' (pid 12345, child pid 12346)
Local: http://localhost:8501
Remote: https://myapp.modelpark.app
Logs: ~/.modelpark/logs/myapp.logOutput (session mode with --session)
Detected framework: streamlit (port 8501)
Registering app 'myapp'...
App 'myapp' registered (private)
Starting: streamlit run app.py
Waiting for port 8501...
Port 8501 is ready
🚀 ModelPark Run
Local: http://localhost:8501
Remote: https://myapp.modelpark.appmodelpark serveTunnel an already-running local port
modelpark lsList running apps
modelpark logsView app logs
modelpark stopGracefully stop an app
modelpark killForce stop an app
modelpark loginAuthenticate with ModelPark
modelpark logoutRemove saved credentials
modelpark statusShow authentication status
modelpark versionPrint version information
Downloads
v2.0.3Released Feb 12, 2026
Release NotesLinux .deb (x86_64)Recommended
4.57 MB · 0 downloads
Linux .rpm (x86_64)Recommended
4.72 MB · 0 downloads
Linux (x86_64)Recommended
4.56 MB · 0 downloads
Linux .deb (ARM64)
4.19 MB · 0 downloads
Linux .rpm (ARM64)
4.29 MB · 0 downloads
Linux (ARM64)
4.18 MB · 0 downloads
macOS (Apple Silicon)
4.35 MB · 0 downloads
macOS (Intel)
4.64 MB · 0 downloads
Windows (x64)
4.7 MB · 0 downloads
Python Library
Prefer Python? Use our Python library to control the ModelPark CLI and access published apps via API.
Features
- Run and serve apps in the background via the CLI
- Manage running apps — list, logs, stop, kill
- Make authenticated API calls to published apps
- API key authentication for REST API apps
- Programmatic access to all ModelPark CLI commands
Installation
pip install modelparkQuick Start — CLI Wrapper
from modelpark import ModelPark
mp = ModelPark()
mp.login(token="your_token")
mp.serve(name="my-app", port=8000)
mp.ls()Quick Start — API Access
from modelpark import APIManager
api = APIManager()
response = api.make_api_call(
"my-app",
headers={"X-ModelPark-Key": "mpk_your_api_key"}
)