ResearchTasksSparse Decoding (GQA)FollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast UpdatedSynthetic 128K context (H100, FP16, page size 16) 1.0 (test)FlashInfer0.19FlashInfer Latency (ms)4May 26, 2026
Synthetic 128K context (H100, FP16, page size 16) 1.0 (test)FlashInfer0.19FlashInfer Latency (ms)4May 26, 2026