Skip to content

aider.md

Latest commit

 

History

History
42 lines (31 loc) · 1.21 KB

aider.md

File metadata and controls

42 lines (31 loc) · 1.21 KB
Raw

Aider + OrcaRouter Lite

Aider speaks any OpenAI-compatible API. Two env vars and you're routing through Lite:

export OPENAI_API_BASE=http://localhost:8000/v1
export OPENAI_API_KEY=sk-orca-PASTE-YOUR-KEY-HERE

# Use any catalog model
aider --model openai/auto                       # ← cheapest capable per request
aider --model openai/gpt-4o
aider --model openai/claude-3-5-sonnet-latest
aider --model openai/gemini-2.5-flash

The openai/ prefix is Aider's way of saying "use the OpenAI-compatible HTTP path." Lite serves that path; the model name after the slash is what gets routed.

Two-model edit/whole pattern

# Cheap model proposes diffs, expensive model confirms them.
aider \
  --model openai/auto \
  --weak-model openai/claude-3-5-haiku-latest

Persistent config

~/.aider.conf.yml:

openai-api-base: http://localhost:8000/v1
openai-api-key: sk-orca-PASTE-YOUR-KEY-HERE
model: openai/auto
weak-model: openai/claude-3-5-haiku-latest

What you get

  • One env var, every model.
  • The local dashboard shows exactly what each Aider session cost.
  • model="auto" makes routine refactors hit Haiku/Flash automatically; complex sessions land on Sonnet/4o.