Same prompt, DeepSeek V4.1 Flash vs Qwen 3.8 FN: 94 tok/s vs 350 tok/s
I wanted to see how big of a difference there is between arguably two of the top MoE models that you can actually run locally, so I ran the same one shot prompt and left it unattended for a while. The prompt asks for a fully interactive V8 engine...