o3

OpenAI · OpenAI o-series · reasoning · released Apr 16, 2025 · o3

OpenAI o-series family:o1·o3·o4-mini

OpenAI's frontier reasoning model — high-compute, extended chain-of-thought for hard math, science, and code.

Pricing

Input$2 / 1M checked Jul 19, 2026
Cached input$2.50 / 1M
Output$8 / 1M checked Jul 19, 2026

Reasoning token usage is included in output cost.

Cost bandPremium
Free tierNo free API tier.

Limits

Context window200k checked Jul 19, 2026
Max output100k
Knowledge cutoffJun 1, 2024
DeploymentOpenAI API
OpenAI-compatible APIYes
ModalitiesText, Image input

Where it fits

Best forOlympiad-level math, advanced coding, PhD-level scientific reasoning where quality trumps cost.
Business fitReserve for the hardest problems; cost is high but so is the ceiling.

Capabilities

Extended thinkingFunction callingStructured outputsVision

Benchmark scores

Composite94 / 100
Coding96 / 100 (SWE-bench Verified)
Reasoning98 / 100 (GPQA Diamond)
Long context90 / 100 (RULER 200K)
Vision88 / 100 (MMMU)
Instruction following94 / 100 (IFEval)

Editorial aggregate of published benchmark results as of May 1, 2025. These are not independent measurements run by this site.

Change history

$0.00$15.33$30.67$46Jun 2026Jun 2026Jul 2026Jul 19, 2026: $40 / 1M to $8 / 1Moutput $8Jul 19, 2026: $10 / 1M to $2 / 1Minput $2

USD per 1M tokens. Steps mark recorded repricings; dots link to the events below.

Data

Last checkedJul 19, 2026
Cross-checked againstopenrouter.ai/openai/o3

Verify with provider documentation before committing spend. Report an error and it gets fixed with a logged correction.

Open in the comparison table JSON API