CoreWeave Cloud pricing

Self-serve published table, North America, per 8-GPU node per hour. On-Demand / Spot / 'Inference Single GPU': HGX H100 $49.24 / $19.71 / $6.16; HGX H200 $50.44 / $20.93 / $6.31; HGX B200 $68.80 / $34.11 / $8.60; HGX A100 80GB $21.60 / $9.65 / $2.70; L40S $18.00 / $7.88 / $2.25; L40 $10.00 / $6.27 / $1.25; GH200 (1 GPU) $6.50 / N/A / $6.50; RTX Pro 6000 Blackwell $20.00 / $11.09 / $2.50. So on-demand is ~$6.16/H100 while spot is ~$2.46/H100 and the inference-platform rate is also $6.16. Europe: H100 $49.24 / $19.51, H200 $50.44 / $20.64, A100 $21.60 / $9.51, B200 $68.80 / $34.87 - spot is regionally cheaper in Europe. GB200 NVL72 is quoted at $42.00/h for the whole 72-GPU Superchip ($10.50 'inference single'), which is far cheaper per GPU than any 8-GPU Blackwell node. Region-restricted deployments carry a stated 1.5x premium. CPU-only nodes also listed.