Use Roo Code with OAIRoute
Roo Code lets every mode use a different model. That only becomes cheap when all those models sit behind one key — which is exactly what the gateway gives you.
Configuration
Provider OpenAI Compatible
Base URL https://api.oairoute.com/v1
API Key YOUR_OAIROUTE_KEY
Model claude-sonnet-5
# Roo supports separate models per mode —
# use a cheaper model for Ask, a stronger one for Code.Why route it through OAIRoute
Ask stays cheap, Code stays smart
Point Ask at a flash-tier model and Code at Sonnet. Typical spend drops by more than half with no loss on the tasks that matter.
One provider entry, many model IDs
You configure OAIRoute once as the provider, then set a different model string per mode.
Same key across every mode
No juggling separate accounts per vendor, no extra billing surfaces to reconcile.
Setup
- 1
Create a key
Register at api.oairoute.com and create a key on the token page.
- 2
Add the provider
In Roo Code settings choose OpenAI Compatible and fill in the base URL and key.
- 3
Assign models per mode
Open each mode's settings and pick its model ID. Ask on a flash model, Code on a Sonnet.
Models worth trying
| Model ID | Name | Use it for | USD per 1M in / out |
|---|---|---|---|
claude-sonnet-5-5 | Claude Sonnet 5.5 | Code mode. Strong at multi-file edits. | $0.18 / $0.90 |
deepseek-v4.1-flash | DeepSeek V4.1 Flash | Ask mode. Cheap enough to leave on. | $0.045 / $0.18 |
claude-opus-5-5 | Claude Opus 5.5 | Architect mode. Deepest reasoning. | $0.20 / $1.00 |
Prices are per million tokens, input / output, at the rates published in the model catalogue.
Questions
Does each mode really need its own entry?
No. Configure the provider once, then set a different model ID inside each mode.
Which mode should get the expensive model?
Architect and Code. Ask and Debug tolerate cheaper models well.
Can I compare two models in the same task?
Yes — same key, different model string. Re-run the task with the other mode pointed at the alternative.