Qwen3.8-Flash-Next (Qwen 3.8 Next Flash) — Windows Vulkan 全适配包 / Windows Vulkan Full Deployment Kit

EN | Complete local deployment kit for the Qwen3.8-Flash-Next 87B MoE (ROCmFP4-FAST-v2-ple16 fork quantization) running on Windows + Vulkan — AMD Ryzen AI MAX+ 395 (ROG Flow Z13, 128GB LPDDR5X, 96GB VRAM carve-out). Includes MTP speculative decoding, PLE n-gram table disk-offload, custom llama.cpp fork binaries, startup scripts and a custom WebUI. This build is original adaptation work — no off-the-shelf guide exists.

中文 | Qwen3.8-Flash-Next 87B MoE(ROCmFP4-FAST-v2-ple16 fork 专用量化)在 Windows + Vulkan 上的完整本地部署方案 — AMD Ryzen AI MAX+ 395(ROG Flow Z13 / 128GB LPDDR5X / 96GB 显存 carve)。含 MTP 投机解码、PLE 表磁盘化、定制 llama.cpp fork 二进制、启动脚本与定制 WebUI。本包为原创适配,网上无现成方案。

🔧 Fork binaries, source patches, scripts & tools → TonyMFJ/Qwen3.8-Flash-Next-WinVulkan 🔧 Fork 二进制、源码补丁、脚本与工具 → TonyMFJ/Qwen3.8-Flash-Next-WinVulkan


📦 Files / 文件清单

File / 文件 Size / 大小 Purpose / 用途
Qwen3.8-Flash-Next-ROCmFP4-FAST-v2-ple16.gguf 87.06 GB Main model, per-head PLE layout (original) / 主模型,per-head PLE 布局(原始版)
ple_sidecar_joined.gguf 20.86 GB Joined-layout PLE table sidecar for --model-ple — disk-offload, frees ~21 GB VRAM / joined 布局 PLE 表侧车,--model-ple 磁盘化加载,腾出约 21GB 显存
mmproj-Qwen3.8-Flash-Next-f16.gguf 0.84 GB Vision projector (must stay on GPU) / 视觉投影器(必须驻留 GPU)
mtp-Qwen3.8-Flash-Next-Q4_K_M.gguf 2.59 GB MTP draft model for speculative decoding (-md) / MTP 草稿模型,投机解码用

💡 Equivalent alternative to the sidecar: convert the main GGUF to joined layout locally with tools/gguf_split_ple_heads.py (join) + --ngram-on-disk. 💡 侧车的等价方案:用 tools/gguf_split_ple_heads.py(join 模式)把主模型本地转成 joined 布局,再配合 --ngram-on-disk

🚀 Key launch flags / 关键启动参数

llama-server -m <ple16.gguf> --model-ple <ple_sidecar_joined.gguf> \
  --mmproj <mmproj-f16.gguf> -md <mtp-Q4_K_M.gguf> \
  -c 102400 --ubatch-size 2048 -ngl 99 --load-mode auto ...
  • EN | --load-mode auto is REQUIRED on this fork/Windows (never none). PLE table on disk ≈ −37% speed but frees ~22 GB VRAM for 100K+ context. Full-VRAM mode reaches ~28 t/s (TG) with MTP acceptance ≈ 67%.
  • 中文 | 本 fork 在 Windows 上必须--load-mode autonone 会崩)。PLE 表磁盘化约损失 37% 速度,换来约 22GB 显存用于 100K+ 上下文。全显存模式 TG 约 28 t/s,MTP 接受率约 67%。

📚 Sources / 来源

📜 Restore walkthrough / 换机恢复指南

See GitHub repo README_恢复指南.md and MODELS.md — unzip, drop models, run one script. 见 GitHub 仓库的 README_恢复指南.mdMODELS.md —— 解压、放模型、跑一个脚本即可。

Downloads last month
470
GGUF
Model size
51B params
Architecture
qwen4exp
Hardware compatibility
Log In to add your hardware

4-bit

Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for TonyMFJ/Qwen3.8-Flash-Next-WinVulkan-GGUF