Qwen3.8-27B — Serving Quantization - a ThakiCloud Collection

ThakiCloud 's Collections

Thaki-HW — Hardware Design LLM (Research Preview)

EmbeddingGemma 2 Edge — Quantization

RAG-Gate — Answer / Retrieve / Stop Gates for RAG

ChainCheck — Chain-Sensitive Evidence Judges

PACE — Enterprise Policy Decisions: Learned or Reasoned

Merchant Desk — Korean Small-Business Customer Desk

Qwen-Image-2.1 — Few-Step Distillation

K-Decision — Korean typed-decision adapters (weights only)

K-MT — Korean-Specialized Translation Without Forgetting

Satoori-KO — Korean Regional Dialect 27B

Evals & Reproducibility

Language Confusion Suppression — Vocabulary Pruning

Vision-Language — Serving Quantization

Qwen3-30B-A3B — Serving Quantization

Qwen3-Coder-30B-A3B — Serving Quantization

Qwen3.8-27B — Serving Quantization

SkillRet — Agent Skill Retrieval

Human-KO — Korean Style-Aligned 27B

Qwen3.8-27B — Serving Quantization

updated Sep 10

Serving-quantization suite for Qwen3.8-27B, all built from the same base so you can pick by hardware generation.

ThakiCloud/Qwen3.8-27B-W4A16-GPTQ

Text Generation • 27B • Updated Sep 7 • 280 • 2

Note W4A16 GPTQ — the default for Ampere and Hopper
ThakiCloud/Qwen3.8-27B-W4A16-AWQ

Text Generation • 27B • Updated Sep 7 • 306 • 1

Note W4A16 AWQ — alternative to GPTQ
ThakiCloud/Qwen3.8-27B-NVFP4-GPTQ-txt

Image-Text-to-Text • 20B • Updated Sep 7 • 217

Note NVFP4 — text-only calibration (Blackwell)
ThakiCloud/Qwen3.8-27B-NVFP4-FP8ATTN

Image-Text-to-Text • 19B • Updated Sep 7 • 100 • 1

Note NVFP4 + FP8 attention, mixed precision