Beam Cloud vs raw cloud GPU pricing - Serverless and On-demand comparison

Fetched 2026-10-04. Beam's on-demand machine rates vs raw cloud providers: H100 PCIe $1.83/hr (Beam) vs AWS p4d.24xlarge $32.77/8 = $4.10/GPU-hr (SkyPilot) vs CoreWeave H100 $6.16/GPU-hr. A100 80GB $1.36/hr (Beam) vs CoreWeave A100 $2.70/GPU-hr. L40S $0.76/hr (Beam) vs CoreWeave L40S $2.25/GPU-hr. RTX 4090 $0.44/hr (Beam) vs Vast.ai median $0.43/GPU-hr. Beam's serverless H100 at $3.50/hr is higher than the on-demand machine rate ($1.83/hr) because serverless includes the per-second billing premium and the orchestration layer. Beam's BYOC model charges $0.019/hr per vCPU + $0.009/hr for management, with compute billed to your own cloud account. The cost calculator on the page shows H100 at $3.50/hr (serverless rate). Beam includes vCPU, RAM, and NVMe in the on-demand machine price, making it an all-in rate. No egress fees on any plan - this is a differentiator vs AWS/GCP/Azure which all charge for data transfer out.