The data-centre Ada card for inference and generative media: 48 GB GDDR6 ECC, FP8 Tensor Cores and AV1 encode, dedicated single-tenant for $0.92/hr in Virginia.
The proven workhorse of AI training and inference, dedicated single-tenant for $1.43/hr in Frankfurt. 80 GB HBM2e at 2 TB/s, MIG, and the most mature software stack in the industry.
The fastest consumer GPU ever made, single-tenant and dedicated for $0.39/hr. 24 GB GDDR6X for 7-13B inference, SDXL, embeddings and rapid experimentation.
Run 70B-class inference, FLUX pipelines and fine-tunes on 96 GB of Blackwell GDDR7 from $0.66/hr - the lowest published rate for this card anywhere.
NVIDIA's flagship Blackwell GPU, on demand from $3.75/hr Dynamic or $6.99/hr Dedicated. 180 GB of HBM3e, 8 TB/s, FP4 Transformer Engine - no waitlist, no 8-GPU minimum.