Did anyone do a full bench of e.g. Qwen Flash Next IQ4 and Qwen 27b FP8? Here are some

A user conducted a quick evaluation of Qwen 3.8 Flash Next IQ4_XS and Qwen 3.8 27B FP8, both served via vLLM. The Qwen 3.8 Flash Next IQ4_XS used r9v, while the Qwen 3.8 27B FP8 utilized Radiance. The evaluation revealed a significant performance difference, with the FP8 27B model running approximately four times faster than the QFN model on the user's system, despite varying concurrency.