VRAM Calculator

Estimate GPU Memory

Calculate VRAM requirements for inference. Adjust model parameters, quantization, and batch size to find the right GPU.

Configuration

512131K
1128

VRAM Breakdown

Weights
7.00 GB
KV Cache
1.07 GB
Overhead
0.64 GB
Total VRAM Required
8.71 GB

GPU Recommendations

GPU VRAM GPUs Utilization $/hour ↑
L424 GB1
36%
$0.48AWS →
RTX 409024 GB1
36%
$0.74RunPod →
A10G24 GB1
36%
$0.75AWS →
A4048 GB1
18%
$0.79Lambda →
L40S48 GB1
18%
$1.14Lambda →
A100 40GB40 GB1
22%
$1.29Lambda →