Aider Multi-Model Configuration

Configures Aider with a cost-optimized multi-model setup using Groq's free-tier LLMs: Llama 3.1 70B for main chat, Mixtral 8x7B for editing tasks, and Llama 3.1 8B for lightweight operations like commit messages.

What This Does

Creates an `.aider.conf.yml` configuration that:

Instructions

1. **Verify Environment**

- Confirm `.env` file exists with `GROQ_API_KEY=<your-key>`

- Aider supports API keys for all providers via `.env` (not just OpenAI/Anthropic in YAML)

2. **Create Configuration File**

- Create `.aider.conf.yml` in project root or home directory

- Copy the configuration below:

```yaml

.aider.conf.yml

Multi-model setup optimized for cost and performance

Main models

model: groq/llama-3.1-70b-versatile

editor-model: groq/mixtral-8x7b-32768

weak-model: groq/llama-3.1-8b-instant

Model settings

architect: true # Use architect mode for complex edits

Cache settings

cache-prompts: true # Reduce API costs

Output settings

dark-mode: true

code-theme: monokai

Git settings

auto-lint: true # Automatically lint after changes

```

3. **Configure Linting (Optional)**

- For Python projects, add: `lint-cmd: "python: flake8 --select=E,W"`

- For JavaScript: `lint-cmd: "javascript: eslint"`

- Or keep `auto-lint: true` and let Aider detect linters

4. **Test Configuration**

```bash

# Start Aider (should load config automatically)

aider

# Verify model in chat

/model

```

5. **Usage Patterns**

- **Complex refactors:** Use `/architect` command or rely on auto-detection

- **Quick edits:** Regular chat uses 70B model efficiently

- **Commit messages:** Auto-generated using 8B model (fast, cheap)

Alternative Model Options

**Replace `model:` line with:**

**For `weak-model:`** (commit messages)

Important Notes

Constraints