R9V Update: now ~100 tok/s in TG on Qwen3.8 Flash Next IQ4_XS on x2 R9700 + 128GB RAM. Fixed crashes with n-gram SSD streaming, improved diagnostics, plus pinned images. Q4_K_XL now supported, 50 tok/s TG.

A new R9V update significantly improves performance and stability, achieving approximately 100 tok/s in TG on Qwen3.8 Flash Next IQ4_XS with x2 R9700 and 128GB RAM. The update fixes crashes related to n-gram SSD streaming, enhances diagnostics, and adds pinned images. It also introduces support for Q4_K_XL, which currently runs at 50 tok/s in TG, with further fine-tuning planned. The developer is also working on an inference engine and a research/site builder application.