Qwen3.8-Flash-Next 125B on AMD R9700

Full local setup and benchmark of the 125B Qwen3.8-Flash-Next model on AMD hardware.

Open source

Qwen3.8-Flash-Next 125B on AMD R9700 media is blocked

Allow external media to connect to the provider and play this content.

Description

A technical demonstration and guide for running the 125-billion-parameter Qwen3.8-Flash-Next Mixture-of-Experts model locally on a single AMD Radeon AI PRO R9700 GPU. The setup utilizes ROCm and llama.cpp to achieve local inference without NVIDIA hardware or cloud-based APIs, proving that flagship-scale models can run on non-CUDA consumer hardware.

Descriptions, tags, and model credits may be AI-generated or inferred from public sources and can be incomplete or wrong. Learn more.

Details

Why this is interesting: Demonstrates the significant capability of running a massive 125B MoE model locally on a single AMD GPU, a major…
Built with: Ling 3.0 Flash Fin Qwen3.8-Flash-Next Qwen3.8-27B Gemini 3.7 Flash Qwen3.7 Flash 3.5 Flash Cyber 3.6 Flash Step 3.7 Flash Gemini 3.5 Flash 2.5 Flash Qwen3 8B Qwen3 14B Qwen3 30B-A3B
Links: View tweet
Tags: #llm #amd-rocm #llama-cpp #local-inference

Discovered: X
First posted: X · Sep 1, 2026
Listed: Added 15d ago
Build: Strong
Models: Strong