@article{latencyprismonlinenonintrusivelatencyscu, title = {LatencyPrism: Online Non-intrusive Latency Sculpting for SLO-Guaranteed LLM Inference}, author = {Yin Du and Jiayi Ren and Xiayu Sun and Tianyao Zhou and Haizhu Zhou and Ruiyan Ma and Danyang Zhang}, year = {2026}, eprint = {2601.09258}, archivePrefix = {arXiv}, url = {https://arxiv.org/abs/2601.09258}, }