o3

OpenAI's most powerful reasoning model

o3 uses extended chain-of-thought reasoning to tackle problems that require deep multi-step thinking: advanced mathematics, competitive programming, PhD-level science, and complex legal analysis.

Try o3

Specifications

Context window200k tokens

Input price$10.00 / 1M tokens

Output price$40.00 / 1M tokens

Credits per query5 cr

Released2025-04

LicenseProprietary

Features

VisionTool useJSON modeCode

Benchmark Scores

What do these mean?

Sourced from official model cards and academic papers. Higher is better.

GPQA Diamond

PhD-level science & reasoning

87.7%

AIME 2024

Mathematical olympiad problems

96.7%

Arena Elo

Chatbot Arena human preference ranking

1420

HumanEval

Python coding — pass@1 rate

99%

MATH

Competition-level mathematics

97.9%

MMLU

General knowledge across 57 academic domains

91.8%

MT-Bench

Multi-turn instruction following (out of 10)

—

SWE-Bench

Real-world GitHub issue resolution rate

71.7%

More from OpenAI

GPT-55 cr

OpenAI's flagship frontier model

GPT-5 Pro5 cr

Highest-capability GPT-5 variant

GPT-5 Mini2 cr

Efficient GPT-5 for everyday tasks