Your AI, Reimagined.

Ever felt that the command line, while powerful, can be a bit... impersonal? Gemini Desktop is a beautiful and intuitive graphical interface for Google's Gemini CLI and Qwen Code, designed to make your interactions with AI more natural and productive.

So, why use a graphical interface instead of the command line?

• It's easier. No need to memorize commands or flags. Everything is a click away.
• It's visual. See your conversations, code changes, and the AI's thought process laid out clearly.
• It's more powerful for complex tasks. Juggling multiple conversations, reviewing large code changes, and managing files is a breeze.
• Access from anywhere. With the web version, you can use your Gemini CLI running on one machine from any other device with a web browser.

Use with Gemini CLI:

Use with Qwen Code:

What you can do with Gemini Desktop

• Chat with your AI, effortlessly. Have natural, flowing conversations with your AI. All your chats are automatically saved for you to come back to at any time.
• See the AI's thought process. Understand how the AI is working behind the scenes by watching its thoughts unfold in real-time.
• Review code changes with confidence. A clear, side-by-side view of proposed code changes makes it easy to see what's new and decide whether to accept it.
• Work with different AI models. Easily switch between different Gemini and Qwen models to find the best one for your needs.
• Search your entire chat history. Instantly find any message or piece of code from your past conversations.

Getting Started

To get started, you'll need to build Gemini Desktop from the source code. Here's how:

1. Install the Prerequisites

Gemini Desktop is built with Rust and web technologies. You'll need to install the following tools first.

• Rust
• Node.js
• pnpm
• just

Installing just

just is a handy command runner we use to simplify the build process.

• On macOS/Linux with asdf:

  asdf plugin add just
  asdf install just latest

• On Ubuntu:

  You can also install just using snap:

  snap install --edge --classic just

• On Windows:

  winget install --id Casey.Just

Linux Dependencies

On Linux, you'll need a few extra packages for the desktop app to build correctly.

• On Ubuntu:

  sudo apt install libgdk-pixbuf-2.0-dev \
     libpango1.0-dev \
     libjavascriptcoregtk-4.1-dev \
     libatk1.0-dev \
     libsoup-3.0-dev \
     libwebkit2gtk-4.1-dev \
     librsvg2-dev

2. Clone the Repository

git clone https://github.com/Piebald-AI/gemini-desktop
cd gemini-desktop

3. Build the Application

This command will download all the necessary dependencies and build both the desktop and web versions of the application.

just deps build-all

4. Run Your Build

Here’s how to run the applications you’ve just built:

Desktop App

For the best experience, use the installer. It will set up the application on your system, just like any other app.

# Navigate to the bundle directory
cd target/release/bundle/

# Run the installer for your system (the file name will vary)
# For example, on Linux it might be:
./Gemini.Desktop_0.1.0_amd64.AppImage

If you prefer to run the application without installing, you can run the executable directly:

./target/release/gemini-desktop

Web App (Web Server)

This will start a web server on your machine. You can then access Gemini Desktop from your browser.

# Run the web server executable
./target/release/gemini-desktop-web

Once the server is running, open your web browser and go to http://localhost:1858.

Features

• Choose between Gemini models (Gemini 2.5 Pro, Gemini 2.5 Flash, Gemini 2.5 Flash-Lite)
• Use Qwen.ai OAuth/custom OpenAI-compatible providers and models with Qwen Code
• Send messages to/from AI and receive response; handle tool call requests; Markdown support
• Observe Gemini's thought process
• View and handle edit requests with clear file diffs
• Automatic light/dark mode switching
• Automatic chat history saving (Note: See related PRs #4401 and #4609 on the Gemini CLI repo.)
• Fast search across all conversation contents
• Access Gemini CLI/Qwen Code running on remote machines with the web UI
• @-mentioning files
• Dangerous --yolo mode support
• MCP server management and integration
• Git repository viewing
• Resuming saved conversations

Planned/in progress

• Token/cost information
• LLxprt integration (#24)
• Multi-modal support
• Extension support

Contributing

Contributions are welcome! Please see the contributing guide for more details.

License

MIT

Copyright © 2025 Piebald LLC.