Open source inference engine tunes itself to your hardware, speeding up open models up to 2x versus llama.cpp on Apple Silicon, NVIDIA, AMD, or CPU-only setups

Original headline: "Open source inference engine (like LM Studio or Unsloth Desktop) that optimizes itself for your exact hardware. Compiles and tunes its kernels on your device, so open models run up to 2x faster than llama.cpp. Works on Apple Silicon, NVIDIA, AMD or nothing but a CPU."