Multi-Model Aider Configuration with OpenRouter & Local Ollama

Configure Aider with a three-tier model strategy: strong reasoning models for complex changes, fast edit models for quick fixes, and lightweight weak models for commit messages. Supports both OpenRouter (free models) and local Ollama deployment.

What This Skill Does

Sets up an `.aider.conf.yaml` file that configures Aider to use:

Instructions

1. **Verify Aider is installed**:

- Check if Aider CLI is available: `aider --version`

- If not installed, install via pip: `pip install aider-chat`

2. **Create the configuration file**:

- Create `.aider.conf.yaml` in your project root

- Copy the multi-model configuration template (see below)

3. **Set up API keys**:

- For OpenRouter models: `export OPENROUTER_API_KEY=your_key_here`

- Get a free API key from https://openrouter.ai/keys

- For local Ollama (optional): `export OLLAMA_API_BASE=http://127.0.0.1:11434`

4. **Choose your model tier**:

- **Free OpenRouter models** (recommended):

- Strong: `openrouter/tngtech/deepseek-r1t2-chimera:free`

- Edit: `openrouter/tngtech/deepseek-r1t2-chimera:free`

- Weak: `openrouter/mistralai/devstral-small-2505:free`

- **Local Ollama models** (for offline use):

- Uncomment the `model: ollama_chat/...` lines

- Start Ollama: `ollama serve`

- Change `openai-api-base` to `http://localhost:11434`

5. **Configure git integration**:

- Auto-commits are **disabled** by default (`auto-commits: false`)

- Create `.aiderignore` file to exclude sensitive files

- Git integration enabled by default (`git: true`)

6. **Alternative model options**:

- **Strong models**: DeepSeek R1 (various versions), OpenRouter Cypher Alpha

- **Edit models**: DeepSeek R1 Qwen3-8B

- **Weak models**: Mistral 7B Instruct, DeepSeek Qwen3-8B

- See commented options in the configuration file

7. **Run Aider with the configuration**:

- Start Aider: `aider`

- Aider will automatically load `.aider.conf.yaml`

- Verify model configuration: Check startup messages

8. **Configuration template**:

```yaml

OpenRouter API endpoint

openai-api-base: https://openrouter.ai/api/v1/chat/completions

Strong model for complex reasoning

model: openrouter/tngtech/deepseek-r1t2-chimera:free

Edit model for quick code changes

editor-model: openrouter/tngtech/deepseek-r1t2-chimera:free

Weak model for commit messages

weak-model: openrouter/mistralai/devstral-small-2505:free

Git settings

git: true

gitignore: false

aiderignore: .aiderignore

auto-commits: false

```

Important Notes

Example Usage

**For a Docker build project**:

```bash

Set API key

export OPENROUTER_API_KEY=sk-or-v1-...

Create configuration

cat > .aider.conf.yaml <<EOF

openai-api-base: https://openrouter.ai/api/v1/chat/completions

model: openrouter/tngtech/deepseek-r1t2-chimera:free

editor-model: openrouter/tngtech/deepseek-r1t2-chimera:free

weak-model: openrouter/mistralai/devstral-small-2505:free

git: true

auto-commits: false

aiderignore: .aiderignore

EOF

Create ignore file

echo "*.env\n.secrets/\n*.key" > .aiderignore

Start Aider

aider

```

**For local Ollama development**:

```bash

Start Ollama

export OLLAMA_API_BASE=http://127.0.0.1:11434

ollama serve &

Pull models

ollama pull qwen2.5:7b

ollama pull gemma3:4b

Update .aider.conf.yaml

Change model to: ollama_chat/qwen2.5:7b

Change openai-api-base to: http://localhost:11434

Start Aider

aider

```

Resources