Welcome to Cognix Discussions!

[cognix-dev]

👋 Welcome!

We're using Discussions as a place to connect with other members of our community. We hope that you:

• Ask questions you're wondering about
• Share ideas and feature requests
• Show off your Cognix workflows
• Help other community members

Quick Links:

• 🐛 Found a bug? → Issues
• 💡 Feature idea? → Start a discussion here!
• 📚 Documentation → README

Let's build amazing AI-powered workflows together! 🚀

[andriy-git]

Awesome! Love the idea. Please make it work with OpenRouter. There are a bunch of models for use and free models as well. I was looking for a cli tool just like this to use with z-ai/glm-4.5-air:free.

[kris3713]

Hey, does this also include Ollama support since your tool claims to also support any OpenAI API compatible backend?

[cognix-dev]

@kris3713 Great question! Cognix uses the OpenAI Python SDK with configurable OPENAI_BASE_URL,
so it can technically connect to any OpenAI-compatible endpoint, including Ollama.

To try it with Ollama:
OPENAI_API_KEY=ollama
OPENAI_BASE_URL=http://localhost:11434/v1

Then:
cognix --model

However, I want to be upfront — Ollama hasn't been formally tested yet,
so there may be edge cases with streaming parameters or model-specific behaviors.
If you try it and run into any issues, please report them here
and I'll work on improving compatibility!

[kris3713]

Thank you for taking the time to reply to my question.

If I run into trouble with Cognix and Ollama, I will let you know when I have the time.

[kris3713]

Hey, I just wanted to let you know that Ollama's OpenAI API works with cognix as seen here with a budget-friendly model (qwen3:0.6b) for demonstration.

Also, I should mention that I no longer use Ollama. I have switched to llama-swap which I found to be way more versatile for my use cases.

[cognix-dev]

@kris3713 , thank you so much for testing and sharing your results!

I'm glad to hear that Cognix can now integrate with the OpenAI-compatible API and that you were able to migrate to llama-swap.

I'm currently working on the next version, which will fully support the /make (build) endpoint for the OpenAI-compatible endpoint. This will allow you to run the entire code generation pipeline, not just chat, in your local LLM.

Please try it out when it's released!