NVIDIA Turing

NVIDIA T4

T4 is a ubiquitous inference GPU ideal for video transcoding and low-cost AI inference deployments.

Launch year
2018
Memory
16 GB GDDR6
Memory bandwidth
320 GB/s
Peak FP16 / FP32
65 TFLOPS · 8 TFLOPS

Market snapshot

$0.07 /hr

Range $0.07 – $13.30

Catalog coverage

6088 live offerings

Across 3 providers · 169 regions

  • 70 W low-profile card
  • Turing Tensor cores
  • Popular in inference clusters
NVIDIA T4 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×0.25
  • Video Transcode×0.40
  • Power Efficiency×0.60

Provider availability

Price bands per provider pulled from the live catalog.

  • Microsoft Azure logoMicrosoft Azure252 offers
    $0.07$7.60 /hr
  • Amazon Web Services logoAmazon Web Services362 offers
    $0.09$13.30 /hr
  • Google Cloud logoGoogle Cloud5474 offers
    $0.09$10.93 /hr

Popular regions

  • europe-west2-a112 offers
  • europe-west2-b112 offers
  • us-west4-b112 offers
  • us-west4-a112 offers
  • asia-southeast1-a112 offers
  • asia-southeast1-b112 offers
  • asia-southeast1-c112 offers
  • asia-east1-a112 offers

Weighted average price $1.47 /hr · median $0.90