List your company

LLM Serving on MATH-plus (p90 latency)

157.83p90 Latency (ms/token)Oracle SJF
138.326269.978401.63533.282Sep 25, 2025

Evaluation results

MethodLinks
Oracle SJFModel=DeepSeek-R1, Workload Mode=Burst Mode2025.09157.83
PARSModel=DeepSeek-R1, Workload Mode=Burst Mode2025.09185.06
Oracle SJFModel=Llama, Workload Mode=Burst Mode2025.09206.64
Listwise SJFModel=DeepSeek-R1, Workload Mode=Burst Mode2025.09208.37
Pointwise SJFModel=DeepSeek-R1, Workload Mode=Burst Mode2025.09231.73
Cross-Model PARSModel=DeepSeek-R1, Workload Mode=Burst Mode2025.09232.14
PARSModel=Llama, Workload Mode=Burst Mode2025.09249.51
Cross-Model PARSModel=Llama, Workload Mode=Burst Mode2025.09281.86
Listwise SJFModel=Llama, Workload Mode=Burst Mode2025.09282.4
Pointwise SJFModel=Llama, Workload Mode=Burst Mode2025.09292.77
FCFSModel=Llama, Workload Mode=Burst Mode2025.09632.7
FCFSModel=DeepSeek-R1, Workload Mode=Burst Mode2025.09645.43