Comparing Qwen 3.6 35B-A3B and Gemma 4 12B for local use

A Reddit user asks for advice on choosing between Qwen 3.6 35B-A3B at Q4 quantization and Gemma 4 12B at Q8 for a 32GB unified memory setup. The user's current Qwen setup runs at ~15 tokens per second. The user wants to compare performance for coding tasks.