Observability & Cost

VRAM-hour billing

VRAM-hour billing charges customers based on the amount of GPU memory (VRAM) they consume over time, rather than a flat per-GPU or per-instance rate.

Instead of paying for a whole GPU regardless of how much of it a workload actually uses, VRAM-hour billing meters and charges for the memory footprint consumed, which can be a fraction of a full GPU's capacity.

Benefits

This aligns cost directly with actual resource consumption, so customers running lightweight or fractional workloads aren't paying full-GPU prices for a sliver of usage.

How hosted·ai approaches this

hosted·ai's pricing model is built around VRAM-consumption-based billing, with no up-front licensing fees and no per-seat fees.

Related terms