Open source · 3 November 2026
Available nowQwenGPU presents Qwen3.5-2B compiled into 45 CUDA-capable GPU layers
qwenGPU compiles a 4-lane SIMT GPU into the Qwen3.5-2B model as 45 untrained layers, enabling the model to generate each step of real CUDA launches using Gated DeltaNet state as registers and memory.
Why it matters
QwenGPU’s ability to generate CUDA code from a 2 B model supports low-latency code-gen features.
For product managers
For your product
What does this mean for your product?
Try it on your product More from this issueAll AI news from 3 Nov →
Get this every morning, written for your product.
Deltaday reads the whole AI news cycle and tells you what changed, whether it's real yet, and what to do about it.