ResearchTasksLLM Serving EfficiencyLLM Serving EfficiencyFollowBenchmarksDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyDatasetSOTA methodMetricTrendResultsLast updatedLMSYS traceToken-budget246GPUs Used2Apr 10, 2026Azure traceHomogeneous358GPUs Used2Apr 10, 2026