Loading page
architecture
Ada Lovelace Architecture GPUs Cloud Pricing
Ada Lovelace is NVIDIA's architecture for the RTX 40-series and professional L-series GPUs. It delivers significant efficiency improvements over Ampere with TSMC 4nm manufacturing and third-generation RT cores. The L40S and L40 serve as cost-effective datacenter options for inference, while RTX 4090 and 4080 are popular for smaller ML workloads.
GPUs 16
Providers 44
From $0.050/hr
Ada Lovelace Architecture GPUs Available in the Cloud
L4
Server24 GBAda LovelaceNVIDIA
$0.130/hr16 providers
L40
40 GBAda LovelaceNVIDIA
$0.330/hr15 providers
L40S
48 GBAda LovelaceNVIDIA
$0.380/hr31 providers
RTX 4000 Ada
20 GBAda LovelaceNVIDIA
$0.160/hr10 providers
RTX 4060
8 GBAda LovelaceNVIDIA
$0.050/hr2 providers
RTX 4060 Ti
8 GBAda LovelaceNVIDIA
$0.071/hr3 providers
RTX 4070
12 GBAda LovelaceNVIDIA
$0.070/hr4 providers
RTX 4070 SUPER
12 GBAda LovelaceNVIDIA
No recent pricing
RTX 4070 Ti
12 GBAda LovelaceNVIDIA
$0.080/hr4 providers
RTX 4070 Ti SUPER
16 GBAda LovelaceNVIDIA
$0.090/hr1 provider
RTX 4080
16 GBAda LovelaceNVIDIA
$0.110/hr4 providers
RTX 4080 SUPER
16 GBAda LovelaceNVIDIA
$0.150/hr4 providers
RTX 4090
24 GBAda LovelaceNVIDIA
$0.160/hr16 providers
RTX 4500 Ada
24 GBAda LovelaceNVIDIA
$0.260/hr1 provider
RTX 5000 Ada
32 GBAda LovelaceNVIDIA
$0.490/hr1 provider
RTX 6000 Ada
48 GBAda LovelaceNVIDIA
$0.390/hr15 providers
Sample Ada Lovelace Architecture GPUs Pricing
| Provider | Price / hr |
|---|---|
$0.080/hr 1× | |
$0.690/hr 2× | |
$0.700/hr 1× | |
$1.00/hr 4× | |
$1.09/hr 6× | |
$1.23/hr 1× | |
$1.57/hr 1× | |
$2.05/hr 1× | |
$2.73/hr 1× |
Direct from providerVia marketplace
Showing 9 of 336 price points. Visit individual GPU pages above for full pricing.
Frequently Asked Questions
Is the RTX 4090 good for machine learning?
The RTX 4090 with 24 GB GDDR6X is effective for inference, fine-tuning, and training smaller models. It lacks the HBM memory and NVLink of datacenter GPUs, so it is not ideal for large-scale distributed training. Its price-to-performance ratio makes it popular for individual researchers.
What is the L40S used for?
The L40S is a datacenter GPU with 48 GB GDDR6 designed for AI inference, video transcoding, and graphics workloads. It supports FP8 for efficient inference and is available across many cloud providers as a mid-range alternative to H100.