NInfer-2080ti-22g now supports Windows-native and is only 0.28 GB
v0.3.0 of NInfer for Turing is out, and this is all about streamlining the UX. For context, this is an engine for running Qwen3.6/8-27B (and ONLY Qwen3.6/8-27B) on a single modded 22 GB RTX 2080 Ti. It is designed to ONLY use VRAM, no system RAM offloading, hence it does not support 35B-A3B.