AWS-optimized GPU for inference, rendering, and video transcoding.
We are tracking a 75% spread for the A10G across 1 cloud provider. The market floor is $1.01/hr per GPU (on-demand), going up to $4.10/hr depending on the provider.
USA 24 configs 1x-8x
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.
24GB GDDR6, AWS-optimized variant of the A10. Good for inference, graphics rendering, and video transcoding on AWS. Widely available on G5 instances.
AWS-only availability. No NVLink, no FP8. For multi-cloud or non-AWS deployments, the standard A10 or L4 is a better fit.
The median on-demand price across providers has risen about 47% since August 2025, from $1.29 to $1.90/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 24GB of VRAM, the A10G can typically run 7B to 13B models in FP16, or larger models in 4-bit quantized form.
The A10G has 24GB of VRAM. Multi-GPU setups increase total memory, but that memory is not automatically pooled across GPUs.
The A10G has 600 GB/s of memory bandwidth. Higher bandwidth helps with faster data transfer between GPU memory and compute cores.
The A10G supports 6 precision formats. Training: BF16, FP16, TF32, FP32. Inference: INT4, INT8.
No. The A10G 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.
A10G pricing currently ranges from $0.43/hr to $4.10/hr per GPU, depending on the provider, instance type, and billing model.
At 720 hours per month, one A10G can cost between $312.91 to $2,949.12 per month, depending on the provider. Reserved and spot pricing can lower that further.
The A10G is available from 1 cloud provider: Amazon Web Services.
Yes. We currently track 24 A10G listings across 1 cloud providers:
| Billing type | Listings | Avg $/GPU/hr |
|---|---|---|
| On-demand | 8 | $1.98/hr |
| Reserved | 16 | $1.22/hr |
| FP32 | 35 TF |
| TF32 Tensor Core | 35 TF |
| BFLOAT16 Tensor Core | 70 TF |
| FP16 Tensor Core | 70 TF |
| INT8 Tensor Core | 140 TOPS |
| INT4 Tensor Core | 280 TOPS |
| RT Cores | 80 |
| Encode / Decode | 1 encoder, 2 decoder (+AV1 decode) |
| GPU Memory | 24 GB GDDR6 |
| GPU Memory Bandwidth | 600 GB/s |
| Interconnect | PCIe Gen4: 64 GB/s |
| Form Factor | 2-slot FHFL |
| Max TDP Power | 300W |
| vGPU Software Support | NVIDIA RTX™ vWS |
Source: official Nvidia A10G datasheet.
Last updated
