T4 variant optimized for AWS Graviton ARM-based instances.
You can start using a T4G for $0.42/hr per GPU (on-demand). It's a low-risk price point for smaller workloads compared to the market high of $1.37/hr.
USA 18 configs 1x-2x
Heads up: We do our best to keep these specs & prices accurate. However, cloud costs may fluctuate based on region, usage, and other factors not listed here. These are estimates based on common setups and are for informational purposes only. Always verify current rates & exact specs with the provider before provisioning.
16GB GDDR6, AWS Graviton-optimized T4 variant. Low-cost inference on ARM-based instances. Good for lightweight models and video processing.
Same Turing limitations as T4. ARM ecosystem may have compatibility gaps. Limited to AWS Graviton instances.
The median on-demand price across providers has risen about 12% since August 2025, from $0.98 to $1.10/hr per GPU. This reflects the market-wide median, which moves both when providers change prices and when lower or higher priced offerings enter the market.
With 16GB of VRAM, the T4G is best for 7B-class models in 4-bit or 8-bit quantized form, and smaller models in FP16.
The T4G has 16GB of VRAM. Multi-GPU setups increase total memory, but that memory is not automatically pooled across GPUs.
The T4G has 320 GB/s of memory bandwidth. Higher bandwidth helps with faster data transfer between GPU memory and compute cores.
The T4G supports 4 precision formats. Training: FP16, FP32. Inference: INT4, INT8.
No. The T4G is a PCIe-only GPU with no NVLink, so it is better suited to single-GPU inference and smaller-scale workloads than large distributed training jobs.
T4G pricing currently ranges from $0.23/hr to $1.37/hr per GPU, depending on the provider, instance type, and billing model.
At 720 hours per month, one T4G can cost between $163.30 to $987.84 per month, depending on the provider. Reserved and spot pricing can lower that further.
The T4G is available from 1 cloud provider: Amazon Web Services.
Yes. We currently track 18 T4G listings across 1 cloud providers:
| Billing type | Listings | Avg $/GPU/hr |
|---|---|---|
| On-demand | 6 | $0.99/hr |
| Reserved | 12 | $0.62/hr |
| FP32 | 8.1 TF |
| FP16 Tensor Core | 16.3 TF |
| INT8 Tensor Core | 32.5 TOPS |
| INT4 Tensor Core | 65 TOPS |
| RT Cores | 40 |
| Encode / Decode | 1 encoder, 2 decoder |
| GPU Memory | 16 GB GDDR6 |
| GPU Memory Bandwidth | 320 GB/s |
| Interconnect | x16 PCIe Gen3 |
| Form Factor | 1-slot Low Profile PCIe |
| Max TDP Power | 70W |
Source: official Nvidia T4G datasheet.
Last updated
