aider is an AI pair programming tool for the terminal. It integrates with Git, supports multi-file editing, and works with any OpenAI-compatible endpoint.
More information: API Access
| Model ID | Best for | Context |
|---|---|---|
qwen3.8-27b |
Code generation, editing, debugging | 256k |
gpt-oss-120b |
Planning and complex reasoning | 128k |
gemma-4-26b-a4b-it |
Quick, lightweight tasks | 256k |
In aider, prefix model IDs with openai/ (for example, openai/qwen3.8-27b).
export OPENAI_API_KEY="your-lsit-api-key"
export OPENAI_API_BASE="https://api.ai.college.ucsb.edu/v1"
Then launch aider with your chosen model:
aider --model openai/qwen3.8-27b
Add the exports to your shell profile (~/.bashrc, ~/.zshrc, etc.) to make them persistent.
.envPlace a .env file in your project root:
OPENAI_API_KEY=your-lsit-api-key
OPENAI_API_BASE=https://api.ai.college.ucsb.edu/v1
Then add a project-level .aider.conf.yml:
model: openai/qwen3.8-27b
Add .env to your .gitignore so the key is never committed.
Architect mode uses one model for planning and a second for editing:
aider --model openai/gpt-oss-120b \
--editor-model openai/qwen3.8-27b \
--architect
Or in .aider.conf.yml:
openai-api-base: https://api.ai.college.ucsb.edu/v1
architect: true
model: openai/gpt-oss-120b
editor-model: openai/qwen3.8-27b
This routes reasoning and task decomposition through gpt-oss-120b and file edits through qwen3.8-27b.
--map-tokens to tune it.--no-auto-commits to disable.