OpenAI
gpt-5.4
Release date not recorded. Accessibility not recorded. Cheapest standard rate we hold: $15 per million output tokens, at short context.
What it costs
| Tier | Kind | In | Out | Cached in | From |
|---|---|---|---|---|---|
| batch | text · long | $2.50 | $11 | $0.25 | seller |
| batch | text · short | $1.25 | $7.50 | $0.13 | seller |
| fast | text · long | $5 | $30 | $0.50 | seller |
| fast | text · short | $5 | $30 | $0.50 | seller |
| flex | text · long | $2.50 | $11 | $0.25 | seller |
| flex | text · short | $1.25 | $7.50 | $0.13 | seller |
| standard | text · long | $5 | $23 | $0.50 | seller |
| standard | text · short | $2.50 | $15 | $0.25 | seller |
per million tokens, US dollars. * marks a rate from a single registry or two that disagree — the best available number rather than a confirmed one. Every other row is confirmed against a seller or a licensed index.
What it scores, by house
This model is not in Epoch's catalogue under a name we can match exactly, so we hold no scores for it. The match is exact after normalisation and never fuzzy, because a loose one would eventually price one model as another.
What retires, and when
- deprecation — 2026-10-01, replaced by gpt-5.6-cyber
Every rate we hold · Why there is no single best model · Everything from OpenAI