Reddit discussion on Gemma4:31b versus Qwen3.6 models

A user reports using Gemma4:31b in opencode for Python scripts, finding it capable but stubborn. Qwen3.6:27b and 35b models are praised for their performance. The thread discusses pragmatic tradeoffs for local coding workloads.