General Tech

Unsloth Qwen3.8-27B Quantizations – Dynamic v3.1 Update

Qwen3.8-27B, a 27B parameter dense model, has been updated with Unsloth Dynamic V3.1, which improves accuracy by 5-10% on most benchmarks. The update includes improved parsing of nested objects, and most quants from 2bit to 6bit have been updated, except for 3bit UD3 quants. The update has been tested on various benchmarks, including Terminal Bench 2.1, DeepSWE v1.1, and LiveCodeBench v6. The updated model can run on 17GB RAM/VRAM setups and is available on Unsloth GGUFs, NVFP4, and Unsloth Desktop.

Read the full article at unsloth.ai →