Switching Between Local and Cloud: Flexibility When You Need It

• Jader Correa

Switching Between Local and Cloud: Flexibility When You Need It

The debate between “Local AI” (privacy, free) and “Cloud AI” (intelligence, speed) is a false dichotomy.

How It Works

Chuchu uses one backend at a time, but you can easily switch between local and cloud configurations depending on your current context:

  • Working on sensitive code? Switch to Ollama (local, private)
  • Need maximum intelligence? Switch to OpenRouter/Groq (cloud, powerful)
  • Offline or traveling? Use Ollama without internet
  • Complex task requiring best models? Use cloud providers

Configuration Setup

Maintain both configurations in your ~/.chuchu/setup.yaml:

defaults:
  backend: ollama  # Default to local

backend:
  ollama:
    type: ollama
    base_url: http://localhost:11434
    default_model: qwen2.5-coder:7b
    agent_models:
      router: llama3.1:8b
      query: qwen2.5-coder:7b
      editor: qwen2.5-coder:7b
      research: llama3.1:8b
  
  groq:
    type: openai
    base_url: https://api.groq.com/openai/v1
    default_model: gpt-oss-120b-128k
    agent_models:
      router: llama-3.1-8b-instant
      query: gpt-oss-120b-128k
      editor: deepseek-r1-distill-qwen-32b
      research: gpt-oss-120b-128k


## Switching Between Backends

### In Neovim

Use the backend selector with `Ctrl+X` in the chat buffer to switch on the fly.

### Via Config File

Edit `~/.chuchu/setup.yaml` and change the `defaults.backend` value:

```yaml
defaults:
  backend: groq  # Switch to cloud

Then restart your chat session.

Benefits of This Approach

  1. Privacy Control: Keep sensitive code local by using Ollama. Switch to cloud only when needed.
  2. Cost Optimization: Use free local models for routine work, pay for cloud only when you need superior intelligence.
  3. Offline Capability: Continue working even without internet by switching to Ollama.
  4. Flexibility: Choose the right tool for the right job - local for speed and privacy, cloud for power.

Hardware Requirements

For effective local model usage:

  • Mac: M1/M2/M3 with at least 16GB RAM.
  • Linux/Windows: NVIDIA GPU with 8GB+ VRAM.

If you don’t have powerful hardware, you can still switch between different cloud providers (Groq for speed, OpenRouter for quality) without running anything locally.

Real-World Usage Patterns

Morning routine work (repetitive, familiar code):

  • Use Ollama locally
  • Fast, free, private
  • Perfect for refactoring, small fixes

Complex feature implementation (new architecture, tricky logic):

  • Switch to OpenRouter with Claude 4.5 Sonnet
  • Maximum intelligence for critical decisions
  • Worth the cost for important work

Quick experiments (learning, trying things out):

  • Use Groq for blazing fast feedback
  • Cheap enough for experimentation
  • Great for iterating quickly

This flexible approach gives you the best of all worlds: privacy when you need it, power when you want it, and cost control always.