List your company
ResearchFormerly WizWand

Scientific Question Answering on GPQA (average inference time (seconds))

1.58Average Inference Time (s)CoT
-3.503630.810765.12599.4393Sep 24, 2025
Updated 7mo ago

Evaluation results

MethodLinks
CoTModel=GPT-3.52025.091.58
Self-reflectionModel=GPT-3.52025.092.44
CoTModel=Mixtral 8x22b2025.093.92
Self-reflectionModel=Mixtral 8x22b2025.096
Self-consistencyModel=GPT-3.52025.096.37
CoTModel=Llama3-70b2025.097.61
CoTModel=GPT-4o-mini2025.097.63
MARSModel=GPT-3.52025.099.54
Self-reflectionModel=GPT-4o-mini2025.0911.45
DMADModel=GPT-3.52025.0912.77
Self-reflectionModel=Llama3-70b2025.0913.24
CoTModel=Mixtral 8x7b2025.0916.05
MADModel=GPT-3.52025.0917.04
MARSModel=Mixtral 8x22b2025.0922.21
Self-consistencyModel=Mixtral 8x22b2025.0923.11
Self-consistencyModel=GPT-4o-mini2025.0929.2
DMADModel=Mixtral 8x22b2025.0933.78
MADModel=Mixtral 8x22b2025.0936.83
Self-reflectionModel=Mixtral 8x7b2025.0938.08
MARSModel=GPT-4o-mini2025.0949.02
MARSModel=Llama3-70b2025.0949.77
Self-consistencyModel=Llama3-70b2025.0952.33
Self-consistencyModel=Mixtral 8x7b2025.0968.17
MARSModel=Mixtral 8x7b2025.0979.24
MADModel=Mixtral 8x7b2025.0979.33
DMADModel=GPT-4o-mini2025.0984.06
MADModel=GPT-4o-mini2025.0985.67
DMADModel=Llama3-70b2025.09102.78
MADModel=Llama3-70b2025.09104.25
DMADModel=Mixtral 8x7b2025.09128.67