Support for other LLM or openrouter

[Hypn0sis]

Hi there, strong project!!! Do you have an roadmap for adding new LLM?

[pratikrath126]

Hi! Many open-source agents support OpenRouter via an OpenAI-compatible endpoint setting (changing the ). Check if Nanoclaw exposes this config. If so, you can point it to OpenRouter and use any LLM.

[joeVenner]

I will work on this update soon this week!

[meglio]

Is it the philosophy of the project to focus on Claude Code only? Or is it just a matter of time before LMStudio, ollama, and other integrations are added?

[joeVenner]

The primary reason is Claude's superior ability to compensate for agent weaknesses. Achieving comparable performance with other models would demand substantial engineering investment.

[anieshchawla]

I replace the Claude in nanoclaw for gemini. As I had only gemini key that has high RPM for model 2.5 flash. https://github.com/anieshchawla/nanoclaw-gemini/tree/main?tab=readme-ov-file

[eli-bar]

Your work to connect NANOCLAW to GEMINI is very impressive. Thank you.
I recently installed NanoClaw V2, and I want to work with Gemini and not with Claud Code.
Do you plan to do it soon with all the changes made in Gemini?
And if not, is it possible to get instructions on how to make this change, what to consider.
I want to develop it in AntiGravity, do you have a Prompt recommendation ?
I would really appreciate it if you could help.
Eli

[burk80]

I am using OpenRouter with the claude api, no problems so far just change the .env anthropic api token

[pjcmyk]

I’d treat endpoint compatibility as a useful first step, while keeping native tool behavior as a separate question. Codelux may help centralize compatible provider and client configuration, but the project still needs to verify its own integration semantics: GitHub - codelux-io/codelux: Unified Provider management CLI for AI coding assistants