L40S vs T4
Lower starting price: T4 ($0.60/GPU-hour difference)
Collected on 2026-10-19
Per GPU-hour (VAT excluded)
Base exchange rate: 1 USD = 1,331 KRW (as of 2026-10-20)
Collected on 2026-10-19View collection & calculation basis →
| Feature | NVIDIA L40S | NVIDIA T4 |
|---|---|---|
| Lowest Price | $1.09 | $0.49₩648 Lower |
| Price Range | $1.09–$3.24 | $0.49–$0.60 |
| Providers | 4 providers | 2 providers |
| VRAM | 48GB | 16GB |
| Form Factor | PCIe | PCIe |
| Use Cases | Training, Inference | Inference |
NVIDIA L40S
| Provider | VRAM | Form Factor | Region | Price / GPU-hr | vs Last Week | Link |
|---|---|---|---|---|---|---|
| RunPod | 48GB | PCIe | Global | $1.09 Lowest | Same | |
| Nebius AI | 48GB | PCIe | Global | $1.55 | Same | |
| CoreWeave | 48GB | PCIe | US | $2.25 | Same | |
| NAVER Cloud | 48GB | PCIe | Korea | $3.24₩4,309 | Same |
NVIDIA T4
| Provider | VRAM | Form Factor | Region | Price / GPU-hr | vs Last Week | Link |
|---|---|---|---|---|---|---|
| Kakao Cloud | 16GB | PCIe | Korea | $0.49₩648 Lowest | Same | |
| NHN Cloud | 16GB | PCIe | Korea | $0.60₩797 | Same |
Sign-up is not open yet.
Model Overview
NVIDIA L40S
A powerful multi-workload data center GPU based on Ada Lovelace architecture with 48GB GDDR6 memory for AI inference and fine-tuning.
NVIDIA T4
Energy-efficient universal GPU based on Turing architecture with 16GB VRAM, widely used for standard ML inference.