GPU Inference Costs

GPU Cost Calculator

Per-camera proportional pricing · pay only for capacity used

Number of Cameras

Hours / Day

Qwen VL Model

Inference Time 10s / chunk

GPU Capacity 6 cameras

Utilization 17%

Customer pays per camera — 1 of 6 slots = 16.7% of GPU cost

☁Cloud GPU

Monthly total · proportional billing

🖥On-Premise

Opex only · one-time capex below

March 2026 · 1 inference/min/camera · 15s chunks · proportional per-camera billing