training

132.65 category score

Model	Precision	Sustained
MobileNetV3	FP32	177.05
MobileNetV3	BF16	122.04
Flan-T5 Small	FP32	192.62
Flan-T5 Small	BF16	108.81
DistilGPT2	FP32	171.26
DistilGPT2	BF16	95.21
DistilBERT	FP32	206.44
DistilBERT	BF16	168.26
GraphGPS (Peptides)	FP32	69.97
GraphGPS (Peptides)	BF16	76.03

inference

235.72 category score

Model	Precision	Sustained
MobileNetV3	FP16	629.11
MobileNetV3	INT8	171.18
Flan-T5 Small	FP16	506.41
Flan-T5 Small	INT8	272.85
Flan-T5 Small	FP8	317.32
DistilGPT2	FP16	278.74
DistilGPT2	INT8	159.97
DistilGPT2	FP8	156.98
DistilBERT	FP16	356.37
DistilBERT	INT8	220.35
DistilBERT	FP8	225.70
GraphGPS (Peptides)	FP16	143.43
GraphGPS (Peptides)	INT8	66.78
GraphGPS (Peptides)	FP8	70.83

compute

61.23 category score

Model	Precision	Sustained
Dense MatMul	FP32	930.27
Dense MatMul	FP16	948.55
Dense MatMul	BF16	952.04
Dense MatMul	FP64	24.74
Sparse MatMul	FP32	6.04
Sparse MatMul	FP64	4.47

Provisional results — not the final, official release · GT = LynxBenchAI Global Score. Every precision result is sustained throughput under a declared, bounded optimisation budget — no collapsed single number below the category level.

← Back to the leaderboard