@article{blinkcpufreellminferencebydelegatingthes,
  title = {Blink: CPU-Free LLM Inference by Delegating the Serving Stack to GPU and SmartNIC},
  author = {Mohammad Siavashi and Mariano Scazzariello and Gerald Q. Maguire and Dejan Kostić and Marco Chiesa},
  year = {2026},
  eprint = {2604.07609},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2604.07609},
}