StartLuxLabs / StartLux-Prism

git clone https://github.com/StartLuxLabs/StartLux-Prism.git

Fast, exact inference for StartLux-Decision models on NVIDIA GPUs: the released GGUF files on int8 and 4-bit tensor cores, behind the /v1/systemone API, with up to 5.3× lower latency than vLLM in FP8 and bit-identical answers.