barozp/Qwen3.8-27B-Opus-Distill-MTP-GGUF overview
Qwen3.8 27B Opus Distill MTP GGUF NOTE Placeholder — not yet available. This repo is reserved ahead of the Qwen/Qwen3.8 27B release and will be filled in once …
Repository Files & Downloads
| File | Type | Quantization | Size | Link |
|---|---|---|---|---|
| Browse files on Hugging Face | ||||
Model Details
Model README
---
license: apache-2.0
tags:
- gguf
- llama.cpp
- qwen
---
Qwen3.8-27B-Opus-Distill-MTP-GGUF
> [!NOTE]
> Placeholder — not yet available. This repo is reserved ahead of the
> Qwen/Qwen3.8-27B release and will be filled in once that model ships.
> Watch this repo or check back after the Qwen3.8 announcement.
Qwen/Qwen3.8-27B fine-tuned on Claude Opus reasoning traces (same recipe as above) with an MTP head grafted on top for self-speculative decoding. The flagship release of this line.
Reasoning-distilled + MTP — the full combination, flagship release.
Related releases in this line
- Qwen3.8-27B-GGUF
- Qwen3.8-27B-MTP-GGUF
- Qwen3.8-27B-Opus-Distill-GGUF
- Qwen3.8-27B-Opus-Distill-MTP-GGUF (this repo)
Prior work from this account
This follows the same recipe (REAP-style efficiency work, Opus reasoning
distillation, MTP grafting, full GGUF quant ladders with measured
speed/quality benchmarks) used for the Qwen3.6 line:
Run barozp/Qwen3.8-27B-Opus-Distill-MTP-GGUF with guIDE
Download guIDE — the AI-native code editor with local LLM inference and 69 built-in tools.
Source: Hugging Face · Compare models