Models/o3

MDL

o3

OpenAI

Verified against Artificial Analysis · Jul 13, 2026

Capability

percentile index

Price /M

$2 / $8

in / out

Context

200K

tokens

Avg score

68.7

11 benchmarks

AA Intelligence

30.4

index

Speed

153 tok/s

TTFT 4.6s

Released

Apr 16, 2025

About

o3 is a proprietary model in the o3 family from OpenAI. Benchmarked on 11 evals, averaging 68.7.

Benchmark scores · 11

avg 68.7 — higher is better

MATHartificial-analysis · 2026-07

99.2

AIMEartificial-analysis · 2026-07

88.3

MMLU-Proartificial-analysis · 2026-07

85.3

GPQA Diamondartificial-analysis · 2026-07

82.7

LiveCodeBenchartificial-analysis · 2026-07

80.8

τ²-benchartificial-analysis · 2026-07

80.7

IFBenchartificial-analysis · 2026-07

71.4

AA Long-Context Reasoningartificial-analysis · 2026-07

69.3

SciCodeartificial-analysis · 2026-07

Terminal-Bench Hardartificial-analysis · 2026-07

37.1

Humanity's Last Examartificial-analysis · 2026-07

Details

PublisherOpenAI

Familyo3

Parameters—

License—

WeightsClosed

ReleasedApr 16, 2025

Performance & price

Output speed153 tok/s

Latency (TTFT)4.6s

Price in / out$2 / $8 per 1M

Blended price$3.5 / 1M

Math index88.3

Median across API providers, via Artificial Analysis. Blended = 3:1 input:output. For reasoning models, latency includes thinking time. Methodology →

History

full ledger →

Capability over time

Append-only ledger — every observed change to this model's numbers.

API providers

Artificial Analysis$8/M