NVIDIA A100 40GB

Cloud GPU Pricing — Live from 8+ Providers
VRAM
40 GB HBM2
Architecture
Ampere
FP16 Performance
312 TFLOPS
Memory Bandwidth
1.6 TB/s
Listings
0 options
Provider Pricing
No pricing data yet — check back soon.
About the NVIDIA A100 40GB

The NVIDIA A100 40GB is ideal for fine-tuning models up to ~30B parameters and running inference on sub-20B models. A cost-effective alternative to the 80GB variant when full VRAM is not required. Popular for Stable Diffusion, image generation, and mid-size LLM workflows.

GPU Specifications
VRAM40 GB HBM2
ArchitectureAmpere
FP16 Performance312 TFLOPS
Memory Bandwidth1.6 TB/s
TDP400W
InterconnectNVLink 3.0
Best For
Model Fine-tuningMid-size Model TrainingBatch InferenceDiffusion Models

Compare All NVIDIA A100 40GB Options Side by Side

Filter by price type, sort by cost, and find the best deal for your workload.

Open Full Comparison →