Ted Hisokawa
Aug 14, 2026 16:26
AMD launches Day 0 support for Qwen3.8 27B AI model on Ryzen AI Max+ processors and Radeon AI PRO R9700 GPUs, enhancing local AI capabilities.
AMD has announced Day 0 support for the newly released Qwen3.8 27B AI model, enabling developers to run the cutting-edge 27-billion-parameter model locally on AMD Ryzen™ AI Max+ processors and AMD Radeon™ AI PRO R9700 GPUs. This marks a significant step in facilitating high-performance local AI applications without reliance on cloud infrastructure.
Qwen3.8 27B, part of Alibaba Cloud’s Qwen family, debuted on August 14, 2026, and is now available on platforms like Hugging Face. It expands on the Qwen3 architecture with enhanced long-context processing and hybrid Gated DeltaNet and Gated Attention layers. The model’s substantial resource requirements—approximately 24GB of VRAM—underscore the importance of robust local hardware for deployment.
On AMD systems, the Qwen3.8 model can be run via open frameworks like llama.cpp, achieving impressive token throughput. Preliminary testing shows performance of up to 24.5 tokens per second on Ryzen AI Max+ 395 processors and 51.8 tokens per second on Radeon AI PRO R9700 GPUs, using Vulkan backend optimizations. These numbers are expected to improve as AMD continues to refine its software stack.
Local AI Simplified with LM Studio
AMD’s LM Studio provides a streamlined interface for running Qwen3.8 27B on supported hardware. Users can download, test, and work with the model directly from a graphical interface, avoiding the need for complex coding. This setup is compatible with Ryzen AI Max+ systems and Radeon GPUs with at least 24GB of VRAM, ensuring a smooth local experience for developers and AI enthusiasts.
Deploying AI with Lemonade
Beyond running the model, AMD’s Lemonade platform simplifies embedding Qwen3.8 27B into local applications. Lemonade acts as a hardware-aware inference layer, optimizing performance across CPUs, GPUs, and NPUs while providing developers with a unified API. This reduces the technical complexity of building AI-driven apps across diverse hardware configurations.
Why This Matters
Qwen3.8 27B represents the next evolution in open-weight AI models, following the Qwen3.6-27B release earlier this year. Its hybrid architecture and long-context capabilities make it well-suited for tasks like coding, research, and extended reasoning, aligning with the broader trend of moving AI workloads from the cloud to local systems.
By delivering Day 0 support, AMD positions itself as a leader in enabling advanced local AI applications, ensuring developers can leverage models like Qwen3.8 immediately upon release. This aligns with the growing demand for decentralized AI solutions, where privacy, latency, and cost concerns drive local deployment.
For developers looking to explore or build on Qwen3.8 27B, AMD’s hardware and software ecosystem—combining Ryzen AI Max+, Radeon GPUs, LM Studio, and Lemonade—offers a compelling foundation to push the boundaries of local AI innovation.
Image source: Shutterstock
Be the first to comment