Longterm Wiki

o1-preview

OpenAI

o1-preview was released September 12, 2024 as OpenAI's first reasoning model. It uses internal chain-of-thought to think through complex problems before responding. Scored 83.3% on the International Mathematics Olympiad qualifying exam (vs 13.4% for GPT-4o) and 78% on GPQA Diamond.

Source →
Developer
OpenAI
Released
2024-09-12
Context Window
128K tokens

Pricing

TypePrice per MTok
Input$15
Output$60

Benchmarks2

BenchmarkScore
GPQA Diamond78%
MATH94.8%

o-series Family5

ModelTierReleasedInput $/MTok
o3flagship2025-04-16$10
o4-minimini2025-04-16$1.1
o3-minimini2025-01-31$1.1
o1flagship2024-12-05$15
o1-minimini2024-09-12$3

Other OpenAI Models7

GPT-4.12025-04-14
GPT (flagship)
GPT-4.1 mini2025-04-14
GPT (mini)
GPT-4.1 nano2025-04-14
GPT (nano)
GPT-4o mini2024-07-18
GPT (mini)
GPT-4o2024-05-13
GPT (flagship)
GPT-4 Turbo2023-11-06
GPT (flagship)
GPT-42023-03-14
GPT (flagship)

Details

Model Familyo-series
Tierpreview
Generationo1
Release Date2024-09-12
Context Window128K tokens

Capabilities1

reasoning

Sources1

Learning to Reason with LLMs

Tags

openaireasoningo-series