Why Chuchu? Building an Affordable AI Coding Assistant

The Problem

AI coding assistants have revolutionized how we write code. Tools like GitHub Copilot, Cursor, and others offer incredible productivity boosts. But there’s a catch: they’re expensive.

GitHub Copilot: $10-20/month
Cursor: $20/month
Other AI IDEs: $15-30/month

For many developers—especially students, open-source contributors, or those in developing countries—these costs add up quickly. A $20/month subscription might not seem like much, but it’s a significant barrier when:

You’re learning to code
You contribute to OSS in your free time
Your local currency makes it 3-5x more expensive
You’re working on personal projects with no income

The Solution: Chuchu

Chuchu is a AI coding assistant with specialized agents and validation designed to be:

1. Radically Affordable

Use the LLM providers you want:

Groq: ~$0.05-0.79 per 1M tokens (blazing fast, ultra-cheap)
Ollama: $0.00 (100% local, completely free)
OpenRouter: Access to 200+ models, pay only for what you use
OpenAI/Anthropic: Direct API access (often 10x cheaper than subscriptions)

Real cost comparison:

Copilot: $20/month = $240/year
Chuchu with Groq: $2-5/month typical usage = $24-60/year
Chuchu with Ollama: $0/year (hardware you already own)

2. Validation ### 2. TDD-First by Design Safety First

Unlike chat-based assistants, Chuchu follows Test-Driven Development:

You: "Add user authentication"
Chuchu: 
  1. Writes failing tests
  2. Implements the feature
  3. Ensures tests pass
  4. Refactors code

This approach:

Catches bugs early
Creates self-documenting code
Builds confidence in AI-generated code
Teaches best practices

3. Neovim Native

Built for Neovim from day one:

Semantic file navigation
Tree-sitter integration
LSP-powered context awareness
Stays out of your way until needed

No Electron. No web UI bloat. Just your terminal and editor.

4. Model Flexibility

Switch between models based on task:

Router agent (fast, cheap): llama-3.1-8b-instant
Query agent (reasoning): gpt-oss-120b or claude-4.5-sonnet
Editor agent (code): deepseek-r1-distill-qwen-32b or qwen-2.5-coder
Research agent (free): grok-4.1-fast (2M context!)

Mix and match. Change anytime. No vendor lock-in.

Who Is This For?

Chuchu is perfect if you:

Can’t afford $20+/month for AI coding tools
Want control over your AI spending
Prefer terminal/Neovim workflows
Care about test coverage
Want to ensure code quality
Need offline/local AI options
Value transparency and open source

The Vision

We believe AI coding assistance should be:

Accessible: No one should be priced out of productivity tools
Transparent: Know what you’re paying, what models you’re using
Educational: Learn best practices while getting work done
Privacy-Conscious: Local models = your code stays yours

Chuchu isn’t trying to replace paid tools for everyone. If Cursor works for you and the cost is fine, that’s great! But for the millions of developers who can’t afford subscriptions, or who want more control, Chuchu offers a real alternative.

Get Started

# Install
go install github.com/jadercorrea/chuchu/cmd/chu@latest

# Configure (use free Groq or local Ollama)
chu setup

# Start coding
nvim

Total cost to try it: $0

Join us in making AI coding assistance accessible to everyone.

Have questions? Join our GitHub Discussions