DeepSeek-V4-Flash-Vision Q8 vs Qwen3.8-Flash-Next Q8

I'm using DS-V4-Flash-Vision with Q8_K_XL quantization locally as my everyday engine, and for some time now I've been doing a lot of comparisons with Qwen3.8-Flash-Next, also with Q8_K_XL quantization. It took me quite a while to get Q3.8FN to work reasonably well, and here are my observations. My h

中文全文翻译正在补齐，可先查看下方英文原文。

I'm using DS-V4-Flash-Vision with Q8_K_XL quantization locally as my everyday engine, and for some time now I've been doing a lot of comparisons with Qwen3.8-Flash-Next, also with Q8_K_XL quantization. It took me quite a while to get Q3.8FN to work reasonably well, and here are my observations. My h