Hacker News Reader

What LLM(s) is the agent using? Are you fine-tuning the model? Is the agent/model development a proprietary effort?

Currently claude 4 sonnet as the main driver, with a combination of smaller models for certain scenarios