Qwen3.8 27B
on NVIDIA RTX 4080 · llama.cpp · 140,000 ctx
Sep 7, 2026
Summary
User reports 8 t/s without MTP, 14 t/s with MTP, and about 19 t/s with full optimizations.
Setup is beellama 0.4.3, a llama.cpp fork.
on NVIDIA RTX 4080 · llama.cpp · 140,000 ctx