Skip to content
Updated minutes ago
nvidia

RTX 3070

nvidia · ampere · 8 GB GDDR6 · 220W TDP

VRAM

8 GB

BF16 TFLOPS

20.4

Bandwidth

448 GB/s

From

$0.09/hr

Calculate ROI with this GPU →

Spec Sheet

VRAM8 GB GDDR6
Memory Bandwidth448 GB/s
BF16 TFLOPS20.4
FP16 TFLOPS20.4
FP8 TFLOPS20.4
INT8 TOPS40.7
TDP220W
InterconnectPCIE
Max per Node4
PCIe Gen4
CUDA Compute Capability8.6
Tensor CoresYes

Pricing by Provider

ProviderOn-DemandReservedSpotBadge
tensordock$0.15/hr-$0.09/hrCheapest
vast_ai$0.18/hr-$0.10/hr

Compatible Models (199)

Training Capabilities

Estimated GPU count for full fine-tuning (AdamW, BF16) and QLoRA

Model SizeFull Fine-TuneQLoRA
7B model17 GPUs1 GPU
13B model31 GPUs1 GPU
70B model165 GPUs6 GPUs

Energy Efficiency

Estimated tokens/second per Watt for popular models

Mistral 7B
0.28 t/s/WFP8
Qwen 2.5 7B
0.27 t/s/WFP8
Llama 3.1 8B
0.25 t/s/WFP8

Similar GPUs

GPUVRAMBF16 TFLOPSBW (GB/s)From
RTX 308010 GB47760$0.14/hr
RTX 306012 GB12.7360$0.06/hr
A400016 GB76448$0.17/hr
A216 GB18200$0.15/hr
A10G24 GB35600$0.30/hr

Methodology Note

Performance estimates for the RTX 3070are based on InferenceBench's roofline performance model with CUDA kernel-level optimization including FlashAttention v2 and PagedAttention. Memory calculations account for model weights (8 GB GDDR6 available), KV-cache allocation, and activation memory. Throughput predictions use the RTX 3070's rated 448 GB/s memory bandwidth and 20.4 BF16 TFLOPS compute capacity as roofline ceilings, with empirical correction factors per GPU architecture (ampere). See our full methodology.

Frequently Asked Questions

How many AI models can run on RTX 3070?

The RTX 3070 can run 199 AI models from our database within a single node. Compatible models range across various parameter sizes depending on the quantization precision (BF16, FP8, INT4). Smaller models fit on a single GPU while larger models may require multi-GPU setups up to 4x RTX 3070.

What is the RTX 3070 inference throughput?

The RTX 3070 delivers 20.4 BF16 TFLOPS and 20.4 FP8 TFLOPS with 448 GB/s memory bandwidth. Actual inference throughput (tokens/sec) depends on the model size, precision, and batch size. Use our calculator for model-specific throughput estimates.

How much does RTX 3070 cost per hour?

The RTX 3070 is available starting from $0.09/hour via tensordock. Prices vary by provider and pricing tier (on-demand, reserved, spot). Compare pricing across all providers in the table above.