ResearchTasksLLM Generation EfficiencyLLM Generation EfficiencyFollowBenchmarksDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyDatasetSOTA methodMetricTrendResultsLast updatedQwen 2.5 14B 2048 input + 8192 generation tokensGRIFFIN287.7End-to-end Latency (s)4Feb 26, 2026Qwen 14B (2048 input + 2048 generation tokens) 2.5GRIFFIN70.3End-to-end Latency (s)4Feb 26, 2026Qwen 2.5 14B (2048 input + 256 generation tokens)GRIFFIN8.7End-to-end Latency (s)4Feb 26, 2026