A developer has compiled a SIMT GPU directly into the weights of Qwen3.5-2B, so the model runs real CUDA kernels inside ...

A developer has compiled a SIMT GPU directly into the weights of Qwen3.5-2B, so the model runs real CUDA kernels inside its own forward passes. The 45 added layers are untrained and built from Qwen's own Gated DeltaNet and SwiGLU blocks, and text drift from the base model stays tiny. Results are self-reported. https://www.marktechpost.com/2026/11/01/qwengpu-simt-gpu-compiled-into-qwen3-5-2b-weights-runs-cuda/ #AIagent #AI #GenAI #AIresearch