Inference
EN
OpenAI

o3

o3 is a reasoning model from OpenAI. 22 benchmarks count toward its index, in 7 categories.

  • Closed weights
  • Reasoning
  • Input: image · text · file
  • Released 2025-04
  • Retires 11 Dec 2026
RankedEnough results for an index.
IndexAll categories, out of 100.51.8±3.9Median 51.0
CoverageIndex weight that has results.95%
SpeedOutput tokens per second.65/sMedian 67/s
First tokenSeconds to the first output token.2.7 sMedian 1.3 s
Input / 1MUS dollars per 1M input tokens.$2.00batch $1.00Median $0.60
Output / 1MUS dollars per 1M output tokens.$8.00batch $4.00 Batch price per 1M output tokens.Median $2.55
ContextMaximum tokens in one request.200KMedian 262K
EloLMArena rating and rank.1410±4 #143Median 1422

Details

From OpenRouter
API IDThe model ID on OpenRouter.
openai/o3
Max outputMaximum output tokens per request.
100K
Knowledge cutoffLast date of training data.
30 Jun 2024
ToolsTool calls on OpenRouter.
Yes
JSON outputOutput follows a given JSON schema.
Yes
Cache read / 1MUS dollars per 1M cached tokens.
$0.50
WeightsLink to the published weights.
—
RetiresDate OpenRouter removes the model.
11 Dec 2026

Where to run

APIs that serve this model.
3 providers
ProviderTypeMaker, cloud, host or router.InputOutputContext
DigitalOceanCloud$2.00$8.00200K
Microsoft AzureCloud$2.00$8.00200K
OpenRouterRouter$2.00$8.00200K

US dollars per 1M tokens, list price. From models.dev (opens in a new tab) (MIT).

Capabilities

Index per category, out of 100.
Out of 100
AgenticMulti-step tasks with tools.1 counted51.8—
CodingCode writing and repair.2 counted51.0131 / 209
ReasoningLogic problems and puzzles.6 counted52.998 / 227
MultimodalTasks with images and text.1 counted53.1—
KnowledgeFacts and expert knowledge.4 counted53.7106 / 268
MultilingualTasks in many languages.None counted——
InstructionTasks with strict prompt rules.1 counted55.7—
MathMath problems.7 counted49.190 / 168

Results

22 counted
BenchmarkThe benchmark name.ResultThe score from the source.PlacePlace among results on this benchmark.
MATH level 5Math97.8% ±0.32 / 88
MATH 500Math94.6% ±1.08 / 59
Artificial Analysis MATH-500Math99.2%2 / 9
Artificial Analysis IFBenchInstruction71.4%38 / 146
SWE-bench Verified · swebench.comCoding58.4%14 / 46
LiveCodeBenchCoding83.9% ±1.044 / 142
SimpleQA VerifiedKnowledge49.4% ±1.631 / 91
GPQA diamond · Epoch AIKnowledge81.8% ±2.182 / 217
GPQA Diamond · Vals AIKnowledge84.1% ±1.952 / 135
AIMEMath85.3% ±1.437 / 95
OTIS Mock AIME 2024-2025Math83.9% ±4.478 / 200
Mystery Game PuzzlesReasoning29.0% ±4.632 / 80
MMLU ProKnowledge85.6% ±0.355 / 135
Artificial Analysis Long Context ReasoningReasoning74.7%97 / 217
τ²-Bench Tool-Agent-User EvaluationAgentic80.7%67 / 149
Artificial Analysis GPQA DiamondKnowledge82.7%91 / 199
FrontierMath-2025-02-28-PrivateMath18.7% ±2.332 / 69
Chess PuzzlesReasoning36.0% ±4.842 / 90
MMMU ProMultimodal80.4% ±1.044 / 92
Artificial Analysis Humanity's Last ExamKnowledge20.1%112 / 222
SWE-bench Verified · SWE-bench/experimentsCoding58.4%33 / 65
ARC-AGI-1 (semi-private)Reasoning60.8%67 / 126
ARC-AGI-2 (semi-private)Reasoning6.5%72 / 131
Critical Physics TasksReasoning1.1%118 / 214
FrontierMath-Tier-4-2025-07-01-PrivateMath2.1% ±2.135 / 55
Artificial Analysis MMMU-ProMultimodal70.1%74 / 113
FrontierMath-Tiers-1-3-v2-PrivateMath33.3% ±2.866 / 90
τ²-bench RetailAgentic73.9%17 / 23
Artificial Analysis MMLU-ProKnowledge85.3%12 / 15
SWE-Bench verified · Epoch AICoding62.3% ±2.227 / 32
Artificial Analysis LiveCodeBenchCoding80.8%11 / 12
τ²-bench AirlineAgentic52.0%21 / 22
MedQAHealthcare96.1% ±0.25 / 95
FrontierMath-Tier-4-2025-07-01-PublicMath0.0%3 / 36
SWE-bench Multimodal · swebench.comMultimodal36.0%1 / 8
SWE-bench Multimodal · SWE-bench/experimentsMultimodal36.0%1 / 7
TaxEval v2Finance74.6% ±0.934 / 143
LegalBenchLegal83.8% ±0.447 / 149
MedCodeHealthcare47.3% ±2.234 / 106
MGSMMultilingual91.7% ±1.726 / 74
MortgageTaxFinance65.7% ±0.940 / 97
FrontierMath-2025-02-28-PublicMath10.0% ±10.029 / 64
ARC-AGI-1 (public eval)Reasoning64.3%63 / 116
CorpFin v2Finance59.7% ±1.074 / 131
ARC-AGI-2 (public eval)Reasoning2.9%80 / 122
MedScribeHealthcare76.7% ±1.974 / 109

Sources

BenchLMCC BY-NC 4.0OpenRouterTerms restrict reuseEpoch AICC BYVals AINo licence statedSierra ResearchMITARC Prize FoundationTerms restrict reuseSWE-bench teamNo licence statedmodels.devMIT

History

5 releases
Index53.3 → 51.8
23 Oct 202612 Nov 2026
Output / 1M$8.00 → $8.00
23 Oct 202612 Nov 2026

Questions

How much does o3 cost?
$2.00 per 1M input tokens, $8.00 per 1M output tokens.
How good is o3?
Its index is 51.8 out of 100. That is place 137 of 283 ranked models.
What is o3 best at?
Knowledge: 53.7 out of 100, place 106 of 268.
How fast is o3?
65 output tokens per second, 2.7 s to the first token.
What is the context window of o3?
200K tokens.
Is o3 open source?
No. Its weights are not published.
When was o3 released?
April 2025.

Badge

o3 on Inference360: place 137 of 283, index 51.8
HTML
<a href="https://inference360.ai/models/o3"><img src="https://inference360.ai/badges/o3.svg" alt="o3 on Inference360: place 137 of 283, index 51.8" height="20"></a>
Markdown
[![o3 on Inference360: place 137 of 283, index 51.8](https://inference360.ai/badges/o3.svg)](https://inference360.ai/models/o3)

Compare with

MiMo-V2-Pro51.8Qwen3.5-122B-A10B51.9GLM-551.6Qwen3.5-27B51.3o4-mini44.7

Lower price

dots3-note Preview63.4 · FreeApodex 1.1 Mini55.7 · FreeNemotron 3 Ultra50.8 · Free

More from OpenAI

All models
o4-mini44.7o4-mini (high)UnrankedGPT-4.138.0GPT-4.1 mini34.5GPT-4.1 nano26.3