NVIDIA
T4
VRAM
16GB
FP32 TFLOPS
8.1 TFLOPS
CUDA Cores
2,560
TDP
70 W
Provider Marketplace
All Cloud Providers
RedSwitchesCheapest
On-Demand•SingaporeSingapore
$0.13/ hour
Estimated Cost
Reserved · commitment
$0.16/ hour
Estimated Cost
On-Demand
$0.20/ hour
Estimated Cost
On-Demand
$0.33/ hour
Estimated Cost
On-Demand
$0.42/ hour
Estimated Cost
On-Demand•China (Beijing),China (Shanghai),Singapore
$2.38/ hour
Estimated Cost
Estimates only — rates are collected automatically from public provider pages and may be out of date. Prices vary by region, commitment term, and availability, and typically exclude storage, egress, and tax. Confirm current pricing with the provider before purchasing. Last collected .
Compute Performance
FP328.1 TFLOPS
FP1665 TFLOPS
INT8130 TOPS
INT4260 TOPS
Architecture
MicroarchitectureTuring
CUDA Cores2560
Tensor CoresTuring Tensor Cores, 320
Dynamic PrecisionSupported (FP32/FP16/INT8/INT4)
Memory & VRAM
Memory TypeGDDR6
Total Capacity16GB
Bandwidth320+ GB/s
ECC SupportYes
Connectivity & Scaling
InterconnectPCIe
GenerationPCIe Gen3
IB Bandwidth32 GB/sec
PCIe InterfaceGen3 xx16
Virtualization
vGPU ReadinessSupported (NVIDIA Virtual Compute Server (vCS))
GPU SharingVirtualization via NVIDIA Virtual Compute Server (vCS)
Power & Efficiency
TDP70 W
Thermal LimitsThermal Solution Passive
Efficiencyproviding an incredible 50X higher energy efficiency compared to CPUs
Physical Design
Form FactorLow-Profile PCIe
CoolingPassive
Rack DensityOptimized for scale-out servers
Thermals & Cooling
DC Heat70 watts
Server & Deployment
Rack-ScaleOptimized for scale-out servers
System Compatibility
Required PCIex16 PCIe Gen3
MotherboardLow-Profile PCIe (small PCIe form factor)
Multi-GPU Scalability
Scaling Efficiency
Single GPU50X higher energy efficiency compared to CPUs
Scaling Characteristics
ParallelismCUDA, NVIDIA TensorRT, ONNX; GPU virtualization via NVIDIA Virtual Compute Server (vCS)
Workload Readiness
Real-Time Serving
up to 40X times better throughput
Market Authority
Key Strengths
Limitations
Also in the Lineup
Expert Insight
The T4 represents a strategic leap in AI compute. When comparing cloud providers, consider not just the hourly rate, but also the interconnect bandwidth (InfiniBand/NVLink) and regional availability which can significantly impact total cost of ownership for large-scale training.
Glossary Terms
FP32 TFLOPS
VRAM
TDP
Cores
Information updated daily. Cloud pricing subject to vendor availability.