Inference
EN
OpenAI

o3-mini

o3-mini is a reasoning model from OpenAI in the o3 family. 23 benchmarks count toward its index, in 6 categories.

  • Closed weights
  • Reasoning
  • Input: text · file
  • Released 2025-01
RankedEnough results for an index.
IndexAll categories, out of 100.40.0±4.9Median 51.0
CoverageIndex weight that has results.85%
SpeedOutput tokens per second.173/sMedian 67/s
First tokenSeconds to the first output token.2.3 sMedian 1.3 s
Input / 1MUS dollars per 1M input tokens.$1.10Median $0.60
Output / 1MUS dollars per 1M output tokens.$4.40Median $2.55
ContextMaximum tokens in one request.200KMedian 262K
EloLMArena rating and rank.1320±4 #240Median 1422

Details

From OpenRouter
API IDThe model ID on OpenRouter.
openai/o3-mini
Max outputMaximum output tokens per request.
100K
Knowledge cutoffLast date of training data.
31 Oct 2023
ToolsTool calls on OpenRouter.
Yes
JSON outputOutput follows a given JSON schema.
Yes
Cache read / 1MUS dollars per 1M cached tokens.
$0.55
WeightsLink to the published weights.
—
RetiresDate OpenRouter removes the model.
—

Capabilities

Index per category, out of 100.
Out of 100
AgenticMulti-step tasks with tools.1 counted21.8—
CodingCode writing and repair.3 counted37.1186 / 209
ReasoningLogic problems and puzzles.4 counted43.2149 / 227
MultimodalTasks with images and text.None counted——
KnowledgeFacts and expert knowledge.5 counted39.0198 / 268
MultilingualTasks in many languages.None counted——
InstructionTasks with strict prompt rules.1 counted52.3—
MathMath problems.9 counted46.1102 / 168

Results

23 counted
BenchmarkThe benchmark name.ResultThe score from the source.PlacePlace among results on this benchmark.
MATH level 5Math96.5% ±0.48 / 88
Instruction-Following EvalInstruction93.9%10 / 70
MATH 500Math91.8% ±2.419 / 59
American Invitational Mathematics Examination 2024Math87.3%9 / 27
SWE-bench Lite · swebench.comCoding32.3%5 / 15
AIMEMath86.5% ±1.132 / 95
Massive Multitask Language UnderstandingKnowledge86.9%13 / 34
SWE-bench Lite · SWE-bench/experimentsCoding31.3%6 / 15
GPQA diamond · Epoch AIKnowledge77.0% ±2.697 / 217
OTIS Mock AIME 2024-2025Math76.9% ±5.292 / 200
FrontierMath-Tier-4-2025-07-01-PrivateMath4.2% ±2.926 / 55
FrontierMath-2025-02-28-PrivateMath12.4% ±1.935 / 69
Graduate-Level Google-Proof Q&AKnowledge77.2%75 / 141
SWE-bench Verified · swebench.comCoding42.4%26 / 46
LiveCodeBenchCoding71.5% ±1.284 / 142
GPQA Diamond · Vals AIKnowledge75.5% ±2.280 / 135
Artificial Analysis GPQA DiamondKnowledge74.8%119 / 199
ARC-AGI-1 (semi-private)Reasoning34.5%86 / 126
ARC-AGI-2 (semi-private)Reasoning3.0%90 / 131
SWE-bench Verified · SWE-bench/experimentsCoding42.4%45 / 65
Artificial Analysis Humanity's Last ExamKnowledge7.9%154 / 222
MMLU ProKnowledge78.7% ±0.4100 / 135
τ²-Bench Tool-Agent-User EvaluationAgentic28.7%114 / 149
Chess PuzzlesReasoning15.0% ±3.673 / 90
Software Engineering Benchmark VerifiedCoding49.3%97 / 114
Mystery Game PuzzlesReasoning7.0% ±2.669 / 80
FrontierMath-Tiers-1-3-v2-PrivateMath18.6% ±2.381 / 90
SimpleQA VerifiedKnowledge15.3% ±1.183 / 91
FrontierMath-Tier-4-v2-PrivateMath0.0%70 / 72
FrontierMath-Tier-4-2025-07-01-PublicMath0.0%3 / 36
MedQAHealthcare94.8% ±0.414 / 95
FrontierMath-2025-02-28-PublicMath40.0% ±16.320 / 64
MGSMMultilingual91.3% ±1.728 / 74
TaxEval v2Finance69.4% ±0.998 / 143
ARC-AGI-1 (public eval)Reasoning46.6%80 / 116
LegalBenchLegal71.5% ±0.7120 / 149
ARC-AGI-2 (public eval)Reasoning0.0%102 / 122
CorpFin v2Finance45.3% ±1.0117 / 131

Sources

BenchLMCC BY-NC 4.0OpenRouterTerms restrict reuseEpoch AICC BYVals AINo licence statedSWE-bench teamNo licence statedARC Prize FoundationTerms restrict reuse

History

5 releases
Index40.9 → 40.0
23 Oct 202612 Nov 2026
Output / 1M$4.40 → $4.40
23 Oct 202612 Nov 2026

Questions

How much does o3-mini cost?
$1.10 per 1M input tokens, $4.40 per 1M output tokens.
How good is o3-mini?
Its index is 40.0 out of 100. That is place 207 of 283 ranked models.
What is o3-mini best at?
Math: 46.1 out of 100, place 102 of 168.
How fast is o3-mini?
173 output tokens per second, 2.3 s to the first token.
What is the context window of o3-mini?
200K tokens.
Is o3-mini open source?
No. Its weights are not published.
When was o3-mini released?
January 2025.

Badge

o3-mini on Inference360: place 207 of 283, index 40.0
HTML
<a href="https://inference360.ai/models/o3-mini"><img src="https://inference360.ai/badges/o3-mini.svg" alt="o3-mini on Inference360: place 207 of 283, index 40.0" height="20"></a>
Markdown
[![o3-mini on Inference360: place 207 of 283, index 40.0](https://inference360.ai/badges/o3-mini.svg)](https://inference360.ai/models/o3-mini)

Compare with

Command A+40.2LFM2.5-24B-A2B39.8Kimi K239.6Grok Code Fast 141.2GPT-4.538.8

Lower price

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

More from OpenAI

All models
o3 Mini HighUnrankedGPT-4.538.8GPT-4o mini AudioUnrankedo1-proUnrankedo144.5