Published pricing (verify on vendor site): On-demand GPU instances billed per hour while running: RTX A6000 48GB $0.35/hr, L40 48GB $0.79/hr, A100 80GB $1.09/hr, H100 SXM 80GB $3.20/hr; extra vCPU $0.04/vCPU-hr; instance disk first 100GB/GPU included then $0.219/GB/mo. Sandbox per-second rates: H100 SXM $2.95/GPU-hr, A100 80GB $1.95/GPU-hr; $0 data egress (per thundercompute.com/pricing).
Updated Oct 25, 2026
Verified-on: 2026-10-25
Launch on-demand GPU instances (A6000, L40, A100, H100) for training or inference via Thunder Compute's redeem_url. Stop or delete instances when idle to avoid charges; confirm current rates at launch.