Why use a router as Aider's backend?

Long edit sessions fire many calls and can trip a provider's rate limit mid-session. CodeBurst fails over across providers behind one model name, so a single provider's limit or outage doesn't interrupt your pair-programming flow.

Guide · Updated June 2026

The best LLM for Aider

Q: How do I use CodeBurst with Aider?

Set OPENAI_API_BASE to https://codeburst.ai/api/v1 and OPENAI_API_KEY to your CodeBurst key, then run aider with the openai/ model prefix: aider --model openai/codeburst-agent.

Q: Do I need the openai/ prefix?

Yes. Aider routes through LiteLLM, so an OpenAI-compatible endpoint needs the openai/ prefix on the model name, e.g. openai/codeburst-agent.

Aider's whole appeal is flow: you stay in the terminal, it edits your files and commits to git while you think. Nothing kills that flow faster than a provider rate-limit error halfway through a refactor. Point Aider at CodeBurst and that single point of failure goes away — one OpenAI-compatible endpoint, automatic failover across providers underneath.

The problem with one provider behind Aider

A focused Aider session is a burst of model calls — diffs, retries, follow-ups — over a short window. That's exactly the traffic shape that trips a single provider's per-minute rate limit. When it does, Aider surfaces the error and your momentum is gone until the window resets. Pinning one provider means your coding session is only as reliable as that provider's worst minute.

What CodeBurst changes

Automatic failover. codeburst-agent is a multi-provider chain — if one provider rate-limits, CodeBurst reroutes to a healthy one in the same request. Aider just keeps going.
Tool-call repair. If a model returns an empty tool-synthesis turn, CodeBurst retries with a corrected format, so edits don't silently stall.
One stable model name. The best underlying model can change without you touching your Aider config.

Set it up

Aider routes through LiteLLM, so an OpenAI-compatible endpoint needs the openai/ prefix on the model name:

export OPENAI_API_BASE="https://codeburst.ai/api/v1"
export OPENAI_API_KEY="YOUR_CODEBURST_KEY"

aider --model openai/codeburst-agent

Prefer a .env file? Use AIDER_OPENAI_API_BASE and OPENAI_API_KEY there, or pass --openai-api-base on the command line. That's the whole change — your workflow is identical, just more resilient.

Get started

Get an API key Best LLM for AI agents →

FAQ

How do I use CodeBurst with Aider?
Set OPENAI_API_BASE=https://codeburst.ai/api/v1 and your key, then aider --model openai/codeburst-agent.

Why a router behind Aider?
Long sessions trip rate limits; CodeBurst fails over across providers so your flow isn't interrupted.

Need the openai/ prefix?
Yes — Aider uses LiteLLM, which needs openai/ for OpenAI-compatible endpoints.