About

Accelerate Qwen3.6 35B inference on RTX 5090 hardware using this hyper-optimized, zero-dependency C/CUDA engine for superior prefill and decode speeds.