ollama-tui

A terminal dashboard for monitoring an Ollama server, built with Rust and Ratatui.

 ollama-tui                    ● Connected  v0.5.1  Refresh: 2s
┌ Running Models (2) ──────────────────────────────────────────┐
│ Name              Size    VRAM   Quant   Ctx    Expires      │
│▶llama3:8b         4.7 GB  100%  Q4_K_M  8192   4m 12s       │
│ mistral:7b        4.1 GB   50%  Q4_0    4096   2m 30s       │
└──────────────────────────────────────────────────────────────┘
┌ Available Models (3) ────────────────────────────────────────┐
│ Name              Size    Family  Params  Quant   Modified   │
│ llama3:8b         4.7 GB  llama   8.0B    Q4_K_M  2d ago     │
│ mistral:7b        4.1 GB  llama   7.2B    Q4_0    5d ago     │
│ gemma3:4b         2.5 GB  gemma3  3.9B    Q4_K_M  12d ago    │
└──────────────────────────────────────────────────────────────┘
 q: Quit  r: Refresh  ↑↓/jk: Scroll  Tab: Switch panel  ?: Help

Features

• Live monitoring -- automatically polls the Ollama API and updates the display
• Running models -- shows loaded models with VRAM/RAM allocation, quantization level, context length, and expiration countdown
• Model catalog -- lists all locally available models with size, family, parameter count, and modification date
• Connection status -- color-coded indicator (green/yellow/red) with server version display
• VRAM color coding -- green for full GPU, yellow for partial offload, red for CPU-only
• Keyboard-driven -- vim-style navigation with scrollable tables and panel switching
• Graceful degradation -- continues polling and displays status when the server is unreachable
• Help overlay -- press ? for a quick keybinding reference

Installation

cargo install --path .

Or build from source:

cargo build --release

The binary will be at target/release/ollama-tui.

Usage

ollama-tui [OPTIONS]

Options

Flag	Default	Description
--host <HOST>	localhost	Ollama server hostname
--port <PORT>	11434	Ollama server port
--refresh <SECS>	2	Polling interval in seconds (minimum: 1)

Examples

# Monitor local Ollama with defaults
ollama-tui

# Connect to a remote server
ollama-tui --host 192.168.1.100

# Custom port and slower polling
ollama-tui --port 8080 --refresh 5

Keybindings

Key	Action
q / Ctrl+C	Quit
r	Refresh immediately
Tab	Next panel
Shift+Tab	Previous panel
j / Down	Scroll down
k / Up	Scroll up
g / Home	Go to top
G / End	Go to bottom
?	Toggle help overlay
Esc	Close help overlay

Architecture

src/
├── main.rs      Entry point, terminal setup, async event loop
├── app.rs       Application state and input handling
├── cli.rs       Command-line argument parsing
├── event.rs     Event multiplexing (keyboard, tick, data updates)
├── ollama.rs    HTTP client and API data structures
├── ui.rs        Ratatui rendering (status bar, tables, help overlay)
└── format.rs    Display formatting (bytes, countdowns, relative dates)

The application uses a background poller task that periodically calls the Ollama API (/api/ps, /api/tags, /api/version) and sends updates through a channel to the main event loop. Terminal input, tick events, and data updates are multiplexed with tokio::select!, keeping the UI responsive while polling runs independently.

Requirements

• Rust 1.86+ (edition 2024)
• A running Ollama server

License

MIT