Updated minutes ago
A2
nvidia · ampere · 16 GB GDDR6 · 60W TDP
VRAM
16 GB
BF16 TFLOPS
18
Bandwidth
200 GB/s
From
$0.15/hr
Spec Sheet
VRAM16 GB GDDR6
Memory Bandwidth200 GB/s
BF16 TFLOPS18
FP16 TFLOPS18
FP8 TFLOPS36
INT8 TOPS36
TDP60W
InterconnectPCIE
Max per Node1
PCIe Gen4
CUDA Compute Capability8.6
Tensor CoresYes
Pricing by Provider
| Provider | On-Demand | Reserved | Spot | Badge |
|---|---|---|---|---|
| retail | $0.15/hr | - | - | Cheapest |
Compatible Models (159)
Single GPU (159 models)
Gemma 2 27B27B INT4Gemma 3 27B27B INT4InternVL2 26B26B INT4Mistral Small 24B24B INT4Mistral Small 3.1 24B24B INT4Codestral 22B22B INT4Solar Pro 22B22B INT4GigaChat 20B20B INT4InternLM 20B20B INT4InternLM 2.5 20B19.9B INT4CogVLM2 19B19B INT4DeepSeek MoE 16B16.4B INT4CodeGen2 16B16B INT4DeepSeek V2 Lite15.7B INT4OctoCoder 15B15.5B INT4StarCoder2 15B15.5B INT4Nemotron 15B15B INT4Qwen 2.5 14B14.8B INT4DeepSeek R1 Distill 14B14.8B INT4Phi-414.7B INT4+139 more
Training Capabilities
Estimated GPU count for full fine-tuning (AdamW, BF16) and QLoRA
| Model Size | Full Fine-Tune | QLoRA |
|---|---|---|
| 7B model | 9 GPUs | 1 GPU |
| 13B model | 16 GPUs | 1 GPU |
| 70B model | 83 GPUs | 3 GPUs |
Energy Efficiency
Estimated tokens/second per Watt for popular models
Mistral 7B
0.46 t/s/WFP8
Qwen 2.5 7B
0.44 t/s/WFP8
Llama 3.1 8B
0.42 t/s/WFP8