OpenAI

davinci-002

Made by OpenAI. Release date not recorded. Accessibility not recorded. Cheapest standard rate we hold: $2 per million output tokens, at long context.

What it costs

TierKindInOutCached inFrom
batchtext$6$6seller
batchtext · long$1$1seller
batchtext · short$1$1seller
standardtext$12$12seller
standardtext · long$2$2seller
standardtext · short$2$2seller

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
metr time horizonscollected 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

  • deprecation2024-01-04, replaced by gpt-3.5-turbo-instruct
  • deprecation2024-01-04, replaced by gpt-3.5-turbo-instruct
  • deprecation2024-10-28, replaced by gpt-4o-mini
  • deprecation2026-09-28, replaced by gpt-5.6-terra
  • deprecation2026-10-23, replaced by gpt-5.6-terra

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