Rent NVIDIA RTX 2070S GPU by the hour
The NVIDIA RTX 2070S is a consumer-segment GPU built on the Turing architecture with 8 GB of VRAM and compute capability 7.5. In ONIE instances, it delivers about 8.8 TFLOPS of standard floating-point performance and about 366 GB/s of memory bandwidth within a 210 W power envelope. You can deploy it with per-hour billing, full SSH access, and ready templates including PyTorch/CUDA, vLLM, Ollama, and ComfyUI.
Specifications
VRAM per GPU
Architecture
Compute capability
FP32 compute
Memory bandwidth
Power limit
Values reported by the infrastructure provider for machines currently listed; real performance varies by machine.
Available configurations
Live stock and prices, updated continuously. The final price is confirmed before you order.
Loading live prices…
What the RTX 2070S is good for
Compact LLM Inference
Run quantized small language models using Ollama or vLLM templates within the 8 GB VRAM capacity.
PyTorch Prototyping and Debugging
Develop and test deep learning code on a Turing GPU with compute capability 7.5 before scaling to larger hardware.
Lightweight Image Generation
Generate images with ComfyUI using standard diffusion pipelines that fit comfortably inside 8 GB of memory.
GPU Compute Workloads
Execute custom CUDA applications utilizing about 8.8 TFLOPS of single-precision compute and about 366 GB/s of memory bandwidth.
Is it the right GPU for you?
Choose the RTX 2070S for development, debugging, and lightweight inference tasks that fit within 8 GB of VRAM. Move to a GPU with more memory and higher compute capability for training large models, or pick a lower-power consumer card for basic script testing.
Compare with similar GPUs
Frequently asked questions
Is the RTX 2070S suitable for LLM inference or fine-tuning?
With 8 GB of VRAM and about 366 GB/s of memory bandwidth, the RTX 2070S handles inference for small, quantized language models via Ollama or vLLM. Full fine-tuning or running larger models requires a GPU with more VRAM.
Can I use this GPU for image generation with ComfyUI?
Yes, ONIE provides a ready ComfyUI template that runs on the RTX 2070S. Its 8 GB of VRAM supports standard diffusion workflows, though heavier pipelines with multiple control nets will benefit from a higher-memory GPU.
How does the Turing architecture affect software compatibility?
The RTX 2070S supports CUDA compute capability 7.5, which is compatible with standard PyTorch/CUDA templates. However, workloads that require newer architectural features found in later generations should be deployed on a newer GPU architecture.
How do I connect and how does billing work on ONIE?
Instances are billed on a per-hour basis for the time your machine is active. You can connect directly via SSH or launch pre-configured templates for PyTorch/CUDA, vLLM, Ollama, and ComfyUI.
Rent the RTX 2070S by the hour
Create an account, choose a configuration and connect over SSH. Billing stops when you delete the GPU.
Rent RTX 2070S