GPU Performance Benchmarks

Evaluate and compare the compute capabilities of the leading cloud GPUs. Make data-driven decisions on infrastructure for your LLM training, fine-tuning, and inference workloads.

Compute Power (FP16/BF16 TFLOPS)

Memory Bandwidth (GB/s)

Complete Hardware Statistics

GPU Model	Architecture	VRAM	Bandwidth	FP16 TFLOPS	INT8 TOPS
NVIDIA H100 (SXM)	Hopper	80 GB HBM3	3,350 GB/s	1,979	3,958
NVIDIA A100	Ampere	80 GB HBM2e	2,039 GB/s	624	1,248
NVIDIA L40S	Ada Lovelace	48 GB GDDR6	864 GB/s	362	733
NVIDIA RTX 4090	Ada Lovelace	24 GB GDDR6X	1,008 GB/s	82.6	660

Don't let hardware limit your potential.

Discover which providers have these models in stock right now.

Browse Providers in Stock