NVIDIA Ampere

NVIDIA A10G

A10G bridges the gap between data center and workstation deployments with Tensor and RT Cores for mixed AI and graphics workloads.

Launch year
2021
Memory
24 GB GDDR6
Memory bandwidth
600 GB/s
Peak FP16 / FP32
125 TFLOPS · 62 TFLOPS

Market snapshot

$0.18 /hr

Range $0.18 – $27.69

Catalog coverage

230 live offerings

Across 1 providers · 17 regions

  • 150 W PCIe card
  • Ampere Tensor + RT Cores
  • Attractive inference fleet pricing
NVIDIA A10G render

Last refreshed Oct 20, 2025, 2:01 AM

Performance snapshot

Normalized versus NVIDIA A100 (=1.0). Values use public reference benchmarks for training and inference workloads.

  • AI Inference (FP16)×0.55
  • AI Training (FP16)×0.45
  • Memory Bandwidth×0.30

Provider availability

Price bands per provider pulled from the live catalog.

  • Amazon Web Services logoAmazon Web Services230 offers
    $0.18 – $27.69 /hr

Popular regions

  • sa-east-116 offers
  • eu-north-116 offers
  • us-east-216 offers
  • us-east-116 offers
  • ap-south-116 offers
  • ap-northeast-216 offers
  • ca-central-116 offers
  • us-west-216 offers

Weighted average price $4.29 /hr · median $2.23