OpenAI

gpt-4-0613

Made by OpenAI. Released 2023-06-13. Accessibility not recorded. Cheapest standard rate we hold: $60 per million output tokens, at long context.

What it costs

TierKindInOutCached inFrom
batchtext · long$15$30seller
batchtext · short$15$30seller
standardtext · long$30$60seller
standardtext · short$30$60seller

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

BenchmarkHouseScoreEvaluatedScaffold
chess puzzlesEpoch AI0.0±0.022026-08-07
gpqa diamondEpoch AI0.3±0.022025-01-27
math level 5near its ceilingEpoch AI0.2±0.012025-01-27
mystery game puzzlesEpoch AI0.1±0.032026-08-28
otis mock aime 2024 2025near its ceilingEpoch AI0.0±0.012025-10-23
bbhcollected by Epoch AI0.8
dtbenchnear its ceilingcollected by Epoch AI0.6
forecastbenchcollected by Epoch AI57.8
gsm8kcollected by Epoch AI0.9
lmcacollected by Epoch AI17.1
metr time horizonscollected by Epoch AI4.0
mmlucollected by Epoch AI0.8
trivia qacollected by Epoch AI0.8
weirdmlcollected by Epoch AI0.1

These scores are not comparable between houses and are not added up. A score is dated by when it was evaluated, not when the model was released, and a score without its scaffold is not a measurement — which is why both are printed.

What retires, and when

  • deprecation2026-10-23
  • deprecation2026-10-23

Every rate we hold · Why there is no single best model · Everything from OpenAI