Token-Burst-Aware Capacity Planning for LLM Inference Services: Request Arrival, Token Demand, and Failure Risk Modeling from BurstGPT Traces. JTIE [Internet]. 2026 Aug. 3 [cited 2026 Sep. 3];5(2):142-64. Available from: https://jtie.stekom.ac.id/index.php/jtie/article/view/565