training

206.19 category score

Model	Precision	Sustained
MobileNetV3	FP32	289.97
MobileNetV3	BF16	181.03
Flan-T5 Small	FP32	268.15
Flan-T5 Small	BF16	163.34
DistilGPT2	FP32	276.90
DistilGPT2	BF16	182.47
DistilBERT	FP32	319.49
DistilBERT	BF16	283.27
GraphGPS (Peptides)	FP32	95.92
GraphGPS (Peptides)	BF16	115.98

inference

290.74 category score

Model	Precision	Sustained
MobileNetV3	FP16	961.23
MobileNetV3	INT8	266.79
Flan-T5 Small	FP16	625.54
Flan-T5 Small	INT8	401.95
Flan-T5 Small	FP8	0.00
DistilGPT2	FP16	474.79
DistilGPT2	INT8	273.58
DistilGPT2	FP8	0.00
DistilBERT	FP16	520.47
DistilBERT	INT8	324.84
DistilBERT	FP8	0.00
GraphGPS (Peptides)	FP16	279.26
GraphGPS (Peptides)	INT8	142.39
GraphGPS (Peptides)	FP8	0.00

compute

87.31 category score

Model	Precision	Sustained
Dense MatMul	FP32	1332.97
Dense MatMul	FP16	1287.14
Dense MatMul	BF16	1299.98
Dense MatMul	FP64	38.22
Sparse MatMul	FP32	8.99
Sparse MatMul	FP64	6.42

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