35B: AMD runs a large model on an NPU through Unsloth Desktop
AMD added XDNA 2 NPU support to Unsloth Desktop, so AI models can run locally without the cloud.
AMD made the AI chip inside a PC run a big program by itself. That means AI on your own computer without the internet. It needs a lot of memory.
Hype check
The noise matches the real change.
Our editorial opinion, based on the sources below. How we score
Your call: hype or real?
Real = it will matter in daily use. Hype = louder than it is. One vote, and it can’t be changed.
Our view and readers’ views are opinions, not statements of fact about a product.
Why it matters
Running models on the eases the load on the CPU and GPU and saves power on laptops.
Skip: news for owners of high-end PCs.
The details
- AMD and Unsloth demonstrated a 35-billion-parameter Qwen 3.6 model running on an NPU.
- It is the first time Unsloth Desktop supports NPUs, and it works with the XDNA 2 architecture.
- The runs use Lemonade and FastFlowLM.
- Full context needs about 28 GB for the NPU, and with Windows' default settings 64 GB of system memory.
- Unsloth Desktop's connects to 100+ applications, including Cline, Open WebUI and Visual Studio Code.
Receipts
One trusted outlet reports it. Not confirmed yet. Open the sources and check us.
Written with the help of AI tools and checked against the sources above. How we work
See a mistake? Tell us and we’ll fix it.