Qwen3.8 27B
on 2× NVIDIA RTX 5070 Ti · llama.cpp · 220,000 ctx
Sep 5, 2026
Summary
User benchmarks Qwen3.8-27B on dual RTX 5070 Ti 16GB cards.
Setup is llama.cpp with the Q4_K_XL quant, MTP3 speculative decoding, and 220k context.
The user also tests vLLM and NInfer, and praises llama.cpp for its simplicity and performance.