Skip to content
Frontier Terminal
Prices

gpt-3.5-turbo-instruct

as of

OpenAI lists gpt-3.5-turbo-instruct at $1.50 per million input tokens and $2.00 per million output tokens. That price has held since at least Aug 1, 2024. OpenAI · model:openai:gpt-3.5-turbo-instruct · 4,096 context

$ per million tokens

6 observations on record. A price holds until the next observation, so the line steps.

input output
$0.00$0.50$1.00$1.50$2.00Nov 7 20:52ZJul 24 17:38ZAug 1 07:23ZSep 22 06:01ZSep 22 12:02Zinputoutput

Point by point

One observation per row: what the page said, when it was read, and the source URL.

Every observation of this SKU with prices and provenance
Observed Event In $/Mtok Cached in Out $/Mtok What changed Source
Internet Archive$1.50—$2.00—web.archive.org
Internet Archive$0.50—$2.00input 1.5 → 0.5web.archive.org
Internet Archive$1.50—$2.00input 0.5 → 1.5web.archive.org
first seen$1.50—$2.00—developers.openai.com
current$1.50—$2.00—developers.openai.com
delisted$1.50—$2.00—developers.openai.com
The 3 “Internet Archive” points come from the Internet Archive’s copy of the lab’s own pricing page, from before we started checking it. Each is the first copy we read that showed that price. We read about one copy a week, so a change can be up to a week earlier than shown. The source link opens that copy. A “first seen” point is the price before our first recorded change. It held from our first check of the page, and any move in between would show up as its own row. A “current” point is the price as of the latest check that changed it.

Data from the last 90 days is CC BY 4.0. Older data is CC BY-NC 4.0, except what we published before 2026-11-04, which stays CC BY 4.0. How you can reuse it.