Ollama CLI TUI

A terminal user interface for managing Ollama models. Built with Python and Textual.

Why This Tool?

The Ollama CLI is powerful but lacks a way to browse and discover models. This TUI solves that by scraping the Ollama registry and providing:

• Full model catalog - Browse 200+ models with descriptions and parameter sizes
• Smart search - Real-time filtering to find the model you need
• Version selection - See all available versions with download sizes before pulling
• One-key actions - Pull, delete, stop models with single keypresses

No more guessing model names or checking the website - everything is accessible from your terminal.

Screenshots

Search & Pull Models

Browse the complete Ollama registry with real-time filtering. See model parameters and descriptions at a glance.

Filter Models

Type to instantly filter the model list. Find what you need in seconds.

Version Selection

When pulling a model, choose the specific version you need. See download sizes before committing.

Manage Local Models

List, inspect, and delete your downloaded models.

Monitor Running Models

See which models are loaded in memory with auto-refresh. Stop them when done.

Features

Model Discovery (Search Tab)

• Browse 200+ models from the Ollama registry
• See parameter sizes (7b, 13b, 70b, etc.) and descriptions
• Real-time filtering - type to search instantly
• Version selection - choose specific tags with download sizes
• Smart caching - 24h cache reduces network requests

Local Model Management (Models Tab)

• List all downloaded models with size and modification date
• View detailed model information (architecture, parameters, license)
• Delete models with confirmation

Process Monitoring (Running Tab)

• Monitor loaded models with GPU/CPU usage
• Auto-refresh every 5 seconds
• Stop running models to free memory

User Experience

• Keyboard-first - Full navigation without mouse
• Visual progress - Progress bar during downloads
• Tab navigation - Switch views with 1/2/3 or arrow keys

Requirements

• Python 3.10+
• Ollama installed and running

Installation

Quick Install (Recommended)

curl -fsSL https://raw.githubusercontent.com/elmisi/ollama-cli-tui/main/install.sh | sh

This clones the repo to ~/.local/share/ollama-tui/, creates a virtual environment, and installs the ollama-tui command to ~/.local/bin/. Running it again will update an existing installation.

Install from Source

git clone https://github.com/elmisi/ollama-cli-tui.git
cd ollama-cli-tui
./install.sh

Uninstall

curl -fsSL https://raw.githubusercontent.com/elmisi/ollama-cli-tui/main/uninstall.sh | sh

Or if installed from a local clone:

./uninstall.sh

Usage

ollama-tui              # Run the TUI
ollama-tui --flush-cache  # Clear cached registry data
ollama-tui --version      # Show version

If running from source:

./run.py
# or
PYTHONPATH=src python -m ollama_tui

Caching

Registry data is cached in ~/.cache/ollama-tui/ for 24 hours to reduce network requests. Use --flush-cache to force a refresh.

Keybindings

Global

Key	Action
1	Switch to Models tab
2	Switch to Running tab
3	Switch to Search tab
← →	Navigate between tabs
q	Quit

Models Tab

Key	Action
↑ ↓	Navigate list
Enter	Show model info
d	Delete model
r	Refresh list

Running Tab

Key	Action
↑ ↓	Navigate list
s	Stop model
r	Refresh list

Search Tab

Key	Action
↑ ↓	Navigate list
/	Focus search input
p	Pull selected model
Enter	Select version (in dialog)
Escape	Back to list / Cancel
r	Refresh from registry

How It Works

Since Ollama doesn't provide a public API for browsing models, this tool scrapes the Ollama Library page to fetch:

• Model names and descriptions
• Available parameter sizes
• Version tags with download sizes

Data is cached locally for 24 hours to be respectful of Ollama's servers.

License

MIT License - see LICENSE for details.

Author

elmisi