Illustrative 10.9 worked example: an hour invoiced at $14 less a $2 failure credit settles at $12; full cost $240 per million accepted tokens
$240 per million accepted tokens (worked example)modeled
| Value kind | modeled — Interpret this value according to its displayed kind, scope, source, and as-of date. |
|---|---|
| Scope | A constructed teaching example: every input (eight GPUs, two on warm standby, $2/active and $1/standby GPU-hour, $12 full hourly cost, $2 credit, 50 of 60 requests accepted) is assumed and stated in Chapter 10.9. It is not a provider quote or a measured result. |
| As of | 2026-09-08 |
| Source | Guide worked example, Chapter 10.9, 8 September 2026; the CloudEvents link documents the event-identity method, not a data source. · Chapter 10.9, worked example: billing an hour with standby and a credit |
| Derivation | 6 GPU-h×$2 + 2 GPU-h×$1 = $14; −$2 credit = $12. 50/60 = 83.33% attainment; 50,000/3,600 = 13.89 tokens/s; $12/(50,000/1e6) = $240/M accepted tokens. If standby goes unbilled at the unchanged $2 active rate, revenue after the credit is $10 against $12 cost; folding standby into the active rate takes $14/6 ≈ $2.33/GPU-h. |
| Review | checking…review by 2027-03-08 · standard cadence |
| Recorded changes | last 2026-10-19 · 2 revisions tracked |
| Claim id | guide2-t210-chargeback |
Where the guide uses it
← Full numbers register — every date-stamped figure in the guide, with revision history.