Qwen3.8 125B (6B active) Flash-Next
on Strix Halo 96GB · llama.cpp · 262,144 ctx
Summary
User is running Qwen Flash Next at IQ4_XS with 262k context on a 96GB Strix Halo. Reports 50 PP and 14 decode near context limit. Seeking speed improvements. Considering V620 32GB via Oculink. Mentions 27B variant as possible alternative.