gemma-4-12b-it vs Qwen3.5-9B on shared benchmarks: Qwen is overall winner beating gemma in 5/8 benchmarks despite a smaller footprint

I don't really understand the gemma hype. Qwen outperforms gemma gb for gb, and kv cache is lighter. Sure gemma-4-12b-it might be a slight better coder than Qwen3.5-9b, but you could also just use omnicoder-9b (Qwen3.5-9