FinOps for LLM inference: unit cost, utilisation, spot and reserved
Compute cost per million tokens from first principles, find where money leaks (idle GPUs, low batch, long prompts, retries), compare on-demand, reserved and spot capacity, build a cost model and dashboard, and a build-versus-buy break-even calculation with worked numbers.
15 min·Verified 2026-11Pro+50 XP
Cost is the metric that turns an inference project from an experiment into a business. This lesson gives you the formulas, the places where money leaks, and a break-even method to decide between self-hosting and an API.
Pro content
Keep reading with Inference Academy Pro
Unlock the rest of this lesson, plus every advanced lesson, hands-on lab, the larger playground models and completion certificates.