Basalt: Fast Blackwell inference engine for Qwen 3.8 Flash Next (and how we got vision for free)

Dynamic vision, in action Basalt is an inference engine for 3.8 Flash Next, forked from Strata and heavily tuned for Blackwell. It supports vision, and for this to work you need to have the image encoder either in VRAM or CPU all the time, having it on VRAM takes valuable space from experts th

中文全文翻译正在补齐，可先查看下方英文原文。

Dynamic vision, in action Basalt is an inference engine for 3.8 Flash Next, forked from Strata and heavily tuned for Blackwell. It supports vision, and for this to work you need to have the image encoder either in VRAM or CPU all the time, having it on VRAM takes valuable space from experts th