BETA

AI-Powered Agentic Coding and More

Desktop app with the heart of a terminal.

Why fight fair? Bring a dragon.

Download for Windows x64 Download for Windows ARM64

· Windows 10/11 · Single executable

Zero Install

Single executable, just download and run

200+ Models

All providers through one interface

Bring Your Subscription

Connect ChatGPT Plus, Gemini Advanced and more

BYOK

Bring your own key, pay only what you use

Local Models

Works offline with Ollama, LM Studio

Keyboard-First

Complete workflows without the mouse

TatsuCode Interface

Slash commands for everything. The speed of CLI with the clarity of GUI.

More Than Chat for LLMs - A Full Environment

File Operations

Read, write, edit, and fuzzy match files project-wide.

Shell Integration

Execute commands with safety controls. No context switching.

Git Aware

View git status and diffs directly from the agent.

Browser Automation

Research, debug, preview and test web apps with internal browser.

SubAgents

Delegate research & heavy tasks to background agents.

Private & Secure

No telemetry. Keys stored locally. You control the code.

The Advantage

Feature TatsuCode Traditional IDEs Web Chat
Installation Zero (Portable) Complex None
Model Access 200+ Models Limited 1 Provider
Offline Capable (Local LLMs) Varies
Browser Agents Limited
Cost Free / BYOK Subscription Subscription

Choose How You Connect

Your Subscriptions

Use your existing subscriptions

  • Google AI Pro
  • ChatGPT Plus
  • More Soon...

API Access

Pay-per-use via OpenRouter

  • 200+ models
  • All major providers
  • One API key

Local Models

Free & offline

  • Ollama
  • LM Studio
  • llama.cpp

Getting Started

Download

Get the single executable

Run

Double-click or run from terminal

Configure

Type /providers to connect

Build

Ship faster with AI assistance

Success is judged by what you ship, not how hard you struggled. AI isn't cheating. It's leverage.
200+ AI Models
14 Themes
0 Telemetry
1 Executable

Frequently Asked Questions

TatsuCode is free to download and use. You only need an API key (OpenRouter) and/or existing subscriptions to OpenAI, Google, or Anthropic. You can also run models locally with Ollama, LM Studio, or llama.cpp — making it possible to use TatsuCode 100% free with local models.
Use the tool that best fits your workflow. TatsuCode offers freedom to choose different models (200+ via OpenRouter), works with existing subscriptions, requires zero installation, and is built natively for Windows. It's also a great "gateway" to agentic programming — if you've never tried these tools, TatsuCode lets you explore without complex setup.
Most agentic tools are built for Mac/Linux first, with Windows as an afterthought. TatsuCode flips that — it's built natively on Windows, for Windows, using .NET. This means better integration, no WSL or Node.js required, and Windows developers aren't second-class citizens.
UI/UX. Image previews, dropdowns, dynamic modals, easier message editing — a GUI offers a better experience while maintaining keyboard-first controls. TatsuCode is a hybrid: desktop chat with terminal-style workflow. ESC closes windows, slash commands for everything.
One sentence: "Reviewing 1,000+ AI-generated pull requests..." No thank you. Quality control and security are priorities. Feature requests are welcome and get implemented when valuable.
Nothing. Zero telemetry. Your chats, sessions, and API keys are stored locally on your machine (keys are encrypted). TatsuCode has no servers collecting your data. For privacy concerns about specific AI models, check the provider's terms directly.
TatsuCode is a solo-developed project. Many weekends and late nights — with many TatsuCode agents helping along the way. This is a passion project built by a developer, for developers.