Updated minutes ago
RTX 3070
nvidia · ampere · 8 GB GDDR6 · 220W TDP
VRAM
8 GB
BF16 TFLOPS
20.4
Bandwidth
448 GB/s
From
$0.09/hr
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
| Provider | On-Demand | Reserved | Spot | Badge |
|---|---|---|---|---|
| tensordock | $0.15/hr | - | $0.09/hr | Cheapest |
| vast_ai | $0.18/hr | - | $0.10/hr |
Compatible Models (181)
Single GPU (69 models)
StarCoder2 7B6.73B FP8DeepSeek Coder 6.7B6.7B FP8MPT 7B6.7B FP8SciGLM 6B6.2B FP8ChatGLM3 6B6B FP8Yi 6B 200K6B FP8Gemma 3 4B4.3B FP8Phi 3.5 Vision4.2B FP8Minitron 4B4B FP8Nemotron Mini 4B4B FP8Qwen 3 4B4B FP8Phi 3 Mini 3.8B3.8B FP8Phi 4 Mini3.8B FP8Stable Diffusion XL 1.03.5B FP8Replit Code v1.5 3B3.3B FP8Llama 3.2 3B3.21B FP8Qwen 2.5 Coder 3B3.1B FP8Qwen 2.5 3B3.09B FP8StarCoder2 3B3.03B FP8OpenELM 3B3B FP8+49 more
Multi-GPU (112 models)
Training Capabilities
Estimated GPU count for full fine-tuning (AdamW, BF16) and QLoRA
| Model Size | Full Fine-Tune | QLoRA |
|---|---|---|
| 7B model | 17 GPUs | 1 GPU |
| 13B model | 31 GPUs | 1 GPU |
| 70B model | 165 GPUs | 6 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