NVIDIA GeForce RTX 4070 Super

NVIDIA12GB GDDR6XPCIe Gen4Ada Lovelace

Data updated 2026-11-04 15:09 UTC · refreshed every 5 min

Current Price

$0.18/hr

From $0.14/hr · lowest live offer

72 GPU · High availability

Current Price

$0.18

-5.1% (24h)

Available Now

72 GPU

High availability

Total in Network

170 GPU

Live data

Utilization

57.6%

Stable

Market history

Price Tracking

USD/hour (minimum · median)

Available GPUs

Ready to rent

Total in Network

Infrastructure Growth

Which LLMs run on one NVIDIA GeForce RTX 4070 Super

Open VRAM calculator with RTX 4070 Super

Estimated with our VRAM calculator for a single card: 8K context, 4 concurrent users, vLLM-style serving.

Largest that fitsINT4 / AWQ

Llama 3.1 8B

8B

9.8 of 12 GB81%

~47 tok/s per stream

Details

Where the memory goes

  • Model weights4.39 GB
  • KV cache4.29 GB
  • Activations0.03 GB
  • CUDA & framework1.04 GB
Precision
INT4 / AWQ
Serving preset
8K context × 4 users
Users at this context
6
Spare context in leftover VRAM
41K
Fit score
62/100 · High
Open in VRAM calculator
BalancedFP8

Qwen3.5-4B

4B

8.2 of 12 GB68%

~56 tok/s per stream

Details

Where the memory goes

  • Model weights6.06 GB
  • KV cache1.07 GB
  • Activations0.03 GB
  • CUDA & framework1.02 GB
Precision
FP8
Serving preset
8K context × 4 users
Users at this context
18
Spare context in leftover VRAM
113K
Fit score
73/100 · Moderate
Open in VRAM calculator
CompactFP16

Qwen3.5-2B

2B

6.1 of 12 GB51%

~74 tok/s per stream

Details

Where the memory goes

  • Model weights4.60 GB
  • KV cache0.40 GB
  • Activations0.05 GB
  • CUDA & framework1.01 GB
Precision
FP16
Serving preset
8K context × 4 users
Users at this context
62
Spare context in leftover VRAM
418K
Fit score
84/100 · Moderate
Open in VRAM calculator

Estimates, not benchmarks: real usage depends on the runtime, context and batch settings.

Technical Specifications

Memory

12GB GDDR6X

FP32 Performance

35.5 TFLOPS

Tensor Performance

568 TFLOPS

Memory Bandwidth

504 GB/s

Architecture

Ada Lovelace

TDP

220W

FP64 Performance

0.55 TFLOPS

CUDA Cores

7168

Tensor Cores

224

Ideal Use Cases

Machine LearningAI Inference3D RenderingGamingContent Creation

About NVIDIA GeForce RTX 4070 Super

The RTX 4070 Super offers excellent efficiency for mid-range AI workloads.

The NVIDIA GeForce RTX 4070 Super features 12GB of GDDR6X memory with 504 GB/s bandwidth, making it excellent for machine learning, ai inference, 3d rendering. With 568 TFLOPS tensor performance and 35.5 TFLOPS FP32 compute, this NVIDIA Ada Lovelace GPU delivers exceptional performance for demanding workloads.

Currently, 170 GPU RTX 4070 Super GPUs are available in the GPUniq network with 57.6% utilization. The average rental price is $0.18 per hour, with 72 GPU ready for immediate deployment.

RTX 4070 Super vs similar GPUs — specs & live rental prices

Prices are live minimum / median USD per hour on the GPUniq marketplace, as of 2026-11-04 15:09 UTC.

GPUVRAMArchitectureFP32TensorBandwidthTDPMin $/hrMedian $/hrAvailable
RTX 4070 Super12GB GDDR6XAda Lovelace35.5 TFLOPS568 TFLOPS504 GB/s220W$0.15$0.2451
RTX 409024GB GDDR6XAda Lovelace82.6 TFLOPS1321 TFLOPS1.01 TB/s450W$0.36$0.81300
RTX 4080 Super16GB GDDR6XAda Lovelace52.2 TFLOPS836 TFLOPS736 GB/s320W$0.21$0.3673
RTX 408016GB GDDR6XAda Lovelace48.7 TFLOPS780 TFLOPS717 GB/s320W$0.25$0.3920

Frequently Asked Questions

How much does it cost to rent NVIDIA GeForce RTX 4070 Super?

The current rental price for NVIDIA GeForce RTX 4070 Super is $0.18 per hour on GPUniq. Prices may vary based on demand and availability.

What is NVIDIA GeForce RTX 4070 Super best used for?

NVIDIA GeForce RTX 4070 Super is ideal for Machine Learning, AI Inference, 3D Rendering, Gaming, Content Creation. With 12GB of GDDR6X memory and 568 TFLOPS tensor performance, it excels at demanding AI and compute workloads.

How many NVIDIA GeForce RTX 4070 Super GPUs are available for rent?

Currently, there are 72 GPU available for immediate rental out of 170 GPU total in the GPUniq network. The current utilization is 57.6%.

Which LLMs can run on NVIDIA GeForce RTX 4070 Super?

On a single NVIDIA GeForce RTX 4070 Super, our VRAM calculator fits Llama 3.1 8B (INT4 / AWQ, 9.8 of 12 GB); Qwen3.5-4B (FP8, 8.2 of 12 GB); Qwen3.5-2B (FP16, 6.1 of 12 GB) — estimated at 8K context with 4 concurrent users.

What are the specifications of NVIDIA GeForce RTX 4070 Super?

NVIDIA GeForce RTX 4070 Super features 12GB GDDR6X memory, 35.5 TFLOPS FP32 performance, 568 TFLOPS tensor performance, 504 GB/s memory bandwidth, and 220W TDP. It uses the Ada Lovelace architecture.

Related GPUs in GeForce RTX 40 Series

GPU rental guides

NVIDIA official specificationsAPI docsAbout GPUniqContact supportPrivacy policyTerms of use

$0.18/hr

RTX 4070 Super · 72 GPU

Rent now