AIMarkTechPostabout 2 hours ago
Developer Releases QwenGPU With CUDA Layers
1 min read
Developer Vixhal released qwenGPU, an experimental model compiling 45 compiled GPU layers into Qwen3.5-2B weights. The 69 total layers run real CUDA on Apple silicon or NVIDIA GPUs requiring 16 GB of free memory. This MIT-licensed tool tests exact computation inside a pretrained LLM.
Level
Hype check
Tap to vote and see what everyone thinks.
