o3

OpenAI's most powerful reasoning model. o3 is a well-rounded and powerful model across domains. It sets a new standard for math, science, coding, and visual reasoning tasks.

COLLIE

98.4%

i
AIME 2024

91.6%

i
ARC-AGI

88.0%

i
MathVista

86.8%

i
AIME 2025

86.4%

i
GPQA

83.3%

i
VideoMMMU

83.3%

i
MMMU

82.9%

i
Aider-Polyglot

81.3%

i
Tau2 Retail

80.2%

i
CharXiv-R

78.6%

i
MMMU-Pro

76.4%

i
SWE-Bench Verified

69.1%

i
Tau2 Airline

64.8%

i
ERQA

64.0%

i
Tau-bench

63.0%

i
Multi-Challenge

60.4%

i
Scale MultiChallenge

60.4%

i
Tau2 Telecom

58.2%

i
Scale MultiChallenge

56.5%

i
BrowseComp

49.7%

i
Humanity's Last Exam

24.3%

i
FrontierMath

15.8%

i
Humanity's Last Exam

14.7%

i
Humanity's Last Exam

14.7%

i
ARC-AGI v2

6.5%

i

Pricing, uptime, and speed via OpenRouter — updated Jul 17, 2026, 04:19 AM.

Provider	Status	Input	Output	Limits	Uptime	Speed	Notes
OpenAI	available	$2.00/Mtok cache $0.50/Mtok	$8.00/Mtok	200K tokens context 100K tokens max output	—	3,017 ms p50 TTFT 79 tok/s p50	$0.01/web search