ResearchBenchmarksComputational Efficiency on Weinreb d2→d6 (inference)Computational Efficiency on Weinreb d2→d6 (inference)Follow1NFE per QueryChreode-2.4821.0144.567.99May 27, 2026Evaluation resultsMethodMethodLinksNFE per QueryLatency (ms) per QueryGFLOPs per QueryChreodeHardware=NVIDIA A100,...Hardware=NVIDIA A100, Precision=fp32, Batch size=12026.05164.790.9CellFlowHardware=NVIDIA A100,...Hardware=NVIDIA A100, Precision=fp32, Batch size=12026.0550410.36—PRESCIENTHardware=NVIDIA A100,...Hardware=NVIDIA A100, Precision=fp32, Batch size=12026.0588194.330.07