CodeMuxWhat is CodeMux?
CodeMux is a multi-model AI coding agent for Visual Studio Code. It can plan, code, and review a task with different models for each step, then presents one result along with the checks it ran.
FAQ
Questions about CodeMux, answered.
Short answers on model routing, verification, supported providers, pricing, and using your own API.
Which AI providers and models does CodeMux support?
The managed catalog currently includes OpenAI: GPT-5.6 Luna, GPT-5.6 Terra, GPT-5.6 Sol; Anthropic: Claude Opus 5, Claude Sonnet 5, Claude Fable 5; DeepSeek: DeepSeek V4 Flash, DeepSeek V4 Pro; xAI: Grok 4.6. Availability can change, so the catalog inside the extension is the authoritative list.
Why use more than one model for a coding task?
Planning, writing code, and reviewing it make different demands on speed, reasoning, and tool use. CodeMux can match each step to a suitable model instead of using the same one for everything.
Does CodeMux verify the code it writes?
It can run your repository’s checks and records the results, including anything it could not verify, before a run is shown as verified. Model output can still be wrong, so you review the changes before keeping them.
How much does CodeMux cost?
The extension is free to install and there is no subscription. Managed models draw from a single pay-as-you-go balance, which you can top up from $5.
Can I use my own AI API?
Yes. Connect a compatible endpoint and CodeMux sends those requests straight to it. Your provider bills you at its own rates; nothing is deducted from your CodeMux balance.
Is CodeMux affiliated with the model providers or other coding agents?
No. CodeMux is an independent product. Third-party company, model, and product names belong to their respective owners.
Still have a question?
Email the CodeMux team. We reply from the same address.