o1

OpenAI's deep thinking model

Last updated: May 22, 2026

OpenAI OpenAI
πŸ“… Released: September 2024 πŸ’³ Paid API DEEP REASONING

Overview

o1 is OpenAI's first-generation reasoning model that spends more time thinking before responding. It excels at breaking down complex problems into steps, verifying its work, and catching its own mistakes. The predecessor to o3, still highly capable for demanding reasoning tasks.

Context Window
200K tokens
Knowledge Cutoff
October 2023
Pricing (Input)
$15.00 / 1M tokens
Pricing (Output)
$60.00 / 1M tokens

βœ…Strengths

  • βœ“Pioneering deep reasoning architecture
  • βœ“Excellent at multi-step problem solving
  • βœ“Self-verification and error correction
  • βœ“Strong performance on STEM benchmarks
  • βœ“200K context for complex analysis

⚠️Weaknesses

  • βœ—Highest latency of all OpenAI models
  • βœ—Very expensive for production use
  • βœ—Older knowledge cutoff than o3/GPT-5
  • βœ—Closed weights - API only

Best Use Cases

πŸ”¬ Research

Complex scientific problems

πŸ“ Mathematics

Proofs, competition problems

πŸ’» Code Review

Security audits, architecture

βš–οΈ Legal

Complex case analysis

πŸŽ“ Education

Advanced tutoring, explanations

πŸ“Š Strategy

Business planning, analysis

Benchmarks

MMLU93.6%
AIME (Math)93.2%
GPQA (Science)92.4%

Other OpenAI Models

πŸš€ Try o1 on OpenAI β†’