⚡ Fast • Lightweight • Terminal

LazyLlama LazyLlama

A blazing fast Terminal User Interface (TUI) client for Ollama with real-time streaming, GitHub-flavored markdown, syntax highlighting, and smart scrolling.

GitHub View on GitHub Crates.io Get from Crates.io
📥 Download LazyLlama
v0.5.3 • GPL-2.0-or-later
lazyllama
$ lazyllama
🦙 LazyLlama v0.5.3 - Starting...
📡 Connected to Ollama
🤖 Available models:
• llama2
• mistral
• codellama
> _

✨ Why LazyLlama?

Built with performance and user experience in mind

⚡

Real-time Streaming

Responses are generated live, providing immediate feedback without waiting for complete generation.

📝

GitHub-Flavored Markdown

Feature-complete markdown rendering with bold, italic, strikethrough, inline code, headers (all 6 levels), ordered/unordered lists, task lists (☐/☑), blockquotes, horizontal rules, and clickable links.

🎨

Syntax Highlighting

Professional-grade code highlighting for 100+ languages (Rust, Python, JavaScript, TypeScript, Go, C++, Java, and more) with 10 customizable themes optimized for both dark and light terminal backgrounds.

🖱️

Smart Scrolling

Automatic following of AI output with manual scroll lock (🔒) for reading previous messages undisturbed.

🔄

Model Management

Easily switch between installed Ollama models using arrow keys with separate input/output buffers per model.

💾

Smart Buffer Management

Each LLM maintains its own chat history, input text, and scroll position for seamless multi-model workflows.

📝

Automatic Logging

Every chat session is automatically saved as text files with both combined and per-model histories.

⚙️

Customizable Settings

Persistent configuration with in-app settings dialog (Ctrl+O). Choose from 10 syntax highlighting themes, with more settings coming soon. Preferences are saved automatically.

⌨️

Emacs-inspired Controls

Familiar keyboard shortcuts for efficient navigation and control, designed for power users.

🦀

Built with Rust

Ultra-low latency and minimal resource footprint using Rust and the Ratatui terminal framework.

⚡ Blazing Fast Performance

Ultra-low latency operations well within 60 FPS frame budget (< 16ms)

LazyLlama is built with performance as a top priority. All operations complete in microseconds or nanoseconds, ensuring a smooth and responsive user experience even with large conversation histories.

🔤 String Operations

Character Insertion 384 ns ⚡⚡
Unicode Handling (🦀) 163 ns ⚡⚡
Pattern Search 7.55 µs ⚡

📝 Text Processing

History Parsing 2.11 µs ⚡
Line Iteration 7.23 µs ⚡
Character Counting 206 ns ⚡⚡

🎯 Cursor Operations

Index Conversion 1 ns ⚡⚡⚡
Word Boundary 2.81 µs ⚡

🌍 Real-World Usage

100KB History 66.2 µs ✅
Model Switching 13.0 µs ✅
HashMap Lookups 6.97 µs ⚡

Performance Legend

⚡⚡⚡ Extreme (< 10 ns)
⚡⚡ Ultra-fast (< 1 µs)
⚡ Fast (< 10 µs)
✅ Good (< 100 µs)

Note: All benchmarks are measured on optimized release builds. Unicode operations are actually faster than ASCII due to efficient UTF-8 handling in Rust. Run cargo test --release --benches to execute the full benchmark suite yourself.

Last measured: 2 March 2026 · Release build · cargo test --release --benches

🚀 Quick Start

Get up and running in minutes

1

Prerequisites

Make sure you have Rust and Ollama installed and running.

2

Install LazyLlama

# From crates.io (recommended)
cargo install lazyllama

# Or build from source
git clone https://github.com/Pommersche92/lazyllama.git
cd lazyllama
cargo install --path .
3

Run LazyLlama

lazyllama

That's it! Start chatting with your AI models right in your terminal.

📥 Alternate download options

Choose your preferred installation method

AppImage

AppImage

Universal Linux package that runs on any distribution

Linux
→
Arch Linux

Arch User Repository

Install directly from AUR using your favorite AUR helper

Arch Linux
→
Windows

Windows Installer

Pre-built binary for Windows 10/11

Windows
→
Crates.io

Cargo Install

Build from source using Rust's package manager

Cross-platform
→
GitHub

Build from Source

Clone the repository and build it yourself

Cross-platform
→

⌨️ Keyboard Controls

Designed for efficiency and productivity

Basic Controls

Enter Send message / Re-activate Autoscroll
Shift + Enter Insert newline (multiline input)
Ctrl + Q Quit application safely
Ctrl + O Open settings dialog (theme selection)
Ctrl + C Clear chat history
Ctrl + S Manually toggle Autoscroll

Clipboard Operations

Ctrl + Shift + C Copy selected text to clipboard
Ctrl + Shift + V Paste text from clipboard

Model & Navigation

Ctrl + ↑ / ↓ Switch between AI Models (loads separate buffers per model)
↑ / ↓ Move cursor up/down between lines in multiline input
PgUp / PgDn Scroll history (activates Manual Mode)
← / → Move cursor left/right in the input field

Text Selection

Shift + ← / → Select text character-wise
Shift + Home / End Select text to start/end of line
Ctrl + Shift + ← / → Select text word-wise

Advanced Editing

Home / End Jump to start/end of the input line
Ctrl + ← / → Move cursor word-wise
Backspace Delete character before the cursor
Delete Delete character after the cursor
Ctrl + Backspace Delete previous word
Ctrl + Delete Delete next word

All keyboard shortcuts are fully customizable via the settings.toml config file.

🔧 How to Configure Keybindings

Create or edit ~/.config/lazyllama/settings.toml (Linux/macOS) or %APPDATA%\lazyllama\settings.toml (Windows) and add a [keybindings] section.

Keybinding format: C- = Ctrl, S- = Shift, C-S- = Ctrl+Shift. Examples:

quit = "C-q"            # Ctrl+Q
copy = "C-S-c"         # Ctrl+Shift+C
insert_newline = "S-Enter"  # Shift+Enter
select_next_model = "C-Down" # Ctrl+Down Arrow
send_query = "Enter"   # Enter (no modifier)

Here's the complete reference of all configurable actions with their defaults:

Application Controls

SettingDefaultAction
quitC-qQuit the application
clear_historyC-cClear current chat history
open_settingsC-oOpen settings dialog
toggle_autoscrollC-sToggle autoscroll mode

Clipboard

SettingDefaultAction
copyC-S-cCopy selected text to clipboard
pasteC-S-vPaste text from clipboard

Input & Cursor Navigation

SettingDefaultAction
send_queryEnterSend message to AI
insert_newlineS-EnterInsert newline in input
cursor_up↑Move cursor up
cursor_down↓Move cursor down
cursor_left←Move cursor left
cursor_right→Move cursor right
cursor_word_leftC-←Jump to previous word
cursor_word_rightC-→Jump to next word
cursor_homeHomeJump to start of input
cursor_endEndJump to end of input

Text Selection

SettingDefaultAction
cursor_left_selectS-←Select text left character-wise
cursor_right_selectS-→Select text right character-wise
cursor_word_left_selectC-S-←Select text left word-wise
cursor_word_right_selectC-S-→Select text right word-wise
cursor_home_selectS-HomeSelect to start of input
cursor_end_selectS-EndSelect to end of input

Model Navigation

SettingDefaultAction
select_next_modelC-↓Switch to next AI model
select_previous_modelC-↑Switch to previous AI model

Deletion

SettingDefaultAction
backspaceBackspaceDelete character before cursor
delete_forwardDeleteDelete character after cursor
delete_word_leftC-BackspaceDelete previous word
delete_word_rightC-DeleteDelete next word

Scrolling

SettingDefaultAction
page_upPageUpScroll history up
page_downPageDownScroll history down

Dialog Keys

These only apply when the settings dialog is open (press Ctrl+O):

SettingDefaultAction
close_dialogEscClose dialog without saving
dialog_up↑Move selection up
dialog_down↓Move selection down
dialog_applyEnterApply selected setting and close

💡 Tip: You only need to specify the keybindings you want to change. Unspecified keybindings use their default values. The status bar in LazyLlama dynamically updates to show your configured keybindings.