Ada Lovelace In stock

NVIDIA RTX 4090

Best raw throughput per dollar on the fleet for single-GPU jobs.

Best suited to

Stable Diffusion
Small model fine-tuning
Local-style dev boxes

Available regions

Region	Location	Typical latency
us-east-1	US East (Ashburn, VA)	12 ms
eu-west-1	EU West (Dublin, IE)	84 ms
ap-south-1	AP South (Mumbai, IN)	146 ms

Hourly rate

$0.44/hr

Billed per second. $10.56 for a full day, $321 if left running a month.

Deploy this GPU All pricing

Full specification

ArchitectureAda Lovelace
VRAM24 GB
Memory typeGDDR6X
Memory bandwidth1.01 TB/s
FP16 dense330 TFLOPS
FP8 dense660 TFLOPS
CUDA cores16,384
Tensor cores512 (4th gen)
Board power450W
InterconnectNot supported
Host vCPU8
Host RAM64 GB
NVMe scratch400 GB
Network100 Gbps Ethernet
MIG partitioningNo

Other Ada Lovelace cards

L40S

48 GB GDDR6 ECC