Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
LLM Serving on MATH-plus (p90 latency)
Loading...
157.83
p90 Latency (ms/token)
Oracle SJF
138.326
269.978
401.63
533.282
Sep 25, 2025
p90 Latency (ms/token)
Updated 26d ago
Evaluation Results
Method
Method
Links
p90 Latency (ms/token)
Oracle SJF
Model=DeepSeek-R1, Wor...
2025.09
157.83
PARS
Model=DeepSeek-R1, Wor...
2025.09
185.06
Oracle SJF
Model=Llama, Workload...
2025.09
206.64
Listwise SJF
Model=DeepSeek-R1, Wor...
2025.09
208.37
Pointwise SJF
Model=DeepSeek-R1, Wor...
2025.09
231.73
Cross-Model PARS
Model=DeepSeek-R1, Wor...
2025.09
232.14
PARS
Model=Llama, Workload...
2025.09
249.51
Cross-Model PARS
Model=Llama, Workload...
2025.09
281.86
Listwise SJF
Model=Llama, Workload...
2025.09
282.4
Pointwise SJF
Model=Llama, Workload...
2025.09
292.77
FCFS
Model=Llama, Workload...
2025.09
632.7
FCFS
Model=DeepSeek-R1, Wor...
2025.09
645.43
Feedback
Search any
task
Search any
task