AMD provides Day 0 support for Alibaba Qwen3.8-27B, runnable on 24GB+ AMD PCs
On August 25, AMD China announced Day 0 support for Qwen3.8-27B, runnable locally on AMD PCs with 24GB+ VRAM.
On August 25, AMD China announced Day 0 support for Alibaba's Qwen3.8-27B. Developers can run the dense model locally on AMD PCs and workstations with 24GB or more VRAM from the model's release day, without waiting for additional compatibility work.
Independent benchmarks show Qwen3.8-27B scoring 52 on the Artificial Analysis Intelligence Index, tying GPT-5.6 Luna. Generation speed sits around 27 tokens/sec, with DFlash2 speculative decoding boosting speed by over 50%. Developers have called it the first mainstream-grade model that fits on a personal computer.
On the same day, the Qwen App and PC client added Alibaba Cloud Token Plan integration. Users can bind an API Key to consume their Token Plan subscription when running complex tasks in Work Assistant, covering computer and browser operations as well as skill-driven delivery of Office documents, apps, and websites.