Qwen3.8 27B
on NVIDIA RTX 5090 · vLLM
Sep 7, 2026
Summary
User reports KLD against BF16 for custom EXL3 quants of Qwen3.8-27B: 0.007406 for EXL3, 0.013126 for FP8, and 0.094978 for NVFP4.
The quants require a custom vLLM fork, Gilded Gnosis.
on NVIDIA RTX 5090 · vLLM