MMLU-ProVerified
85.3%
MMLU-Pro (AA run)
Models / OpenAI
OpenAI o3
OpenAI o · Released 2025-04-16
OpenAI's April 2025 reasoning model. Extended thinking for math, code, and science. LiveBench 68.5%, LongBench v2 51.2%, BFCL 78.1%.
Reasoning-tier OpenAI pick when latency and cost are secondary to hardest single-shot problem solving.
#125 of 168 on AA Index · current snapshot
31 AA Index · 20 benchmark rows · 7 sourcesPosition is benchmark-specific — not a cross-family or cross-source ranking.Composite70
Context200K
Input / 1M—
Output / 1M—
Knowledge cutoff2024-06-01
Output speed134 t/s
TTFT (default API)5.65s
Statuslightweight
Family profile
Best published score in each covered benchmark family.
79/100 avg
- Knowledge
- 85
- Reasoning
- 88
- Math
- 99
- Coding
- 95
- Agentic
- 37
- Long context
- 73
- Tool use
- 78
7 tested benchmark families
Benchmark placements
Per-benchmark positions
Where OpenAI o3 places on each public benchmark source that publishes a row. Rank counts every model with a latest row in the same test — not a universal quality score.
| Benchmark | Family | Rank | Score | Source | Date |
|---|---|---|---|---|---|
| MATH | Math | #2/ 25 | 99 | Artificial Analysis | 2026-09-01 |
| LiveCodeBench | Coding | #17/ 31 | 81 | Artificial Analysis | 2026-09-01 |
| AIME | Math | #20/ 37 | 88 | Artificial Analysis | 2026-09-01 |
| AA time to first token | Performance | #21/ 186 | Artificial Analysis | 2026-09-01 | |
| AA output speed | Performance | #22/ 187 | Artificial Analysis | 2026-09-01 | |
| MMLU-Pro | Knowledge | #22/ 39 | 85 | Artificial Analysis | 2026-09-01 |
| IFBench | Reasoning | #46/ 127 | 71 | Artificial Analysis | 2026-09-01 |
| AA-LCR | Long context | #62/ 179 | 73 | Artificial Analysis | 2026-09-01 |
| SciCode | Coding | #116/ 180 | 41 | Artificial Analysis | 2026-09-01 |
| Humanity's Last Exam | Reasoning | #120/ 180 | 20 | Artificial Analysis | 2026-09-01 |
| GPQA Diamond | Reasoning | #124/ 185 | 83 | Artificial Analysis | 2026-09-01 |
| Terminal-Bench | Agentic | #126/ 184 | 37 | Artificial Analysis | 2026-09-01 |
| Artificial Analysis Intelligence Index | Reasoning | #143/ 187 | 31 | Artificial Analysis | 2026-09-01 |
Family context
Strongest rows in each family
The three highest-scoring models with pages in each capability family. Where OpenAI o3 shows up, it's highlighted.
KnowledgeOpenAI o3 · 85
- 191
- 291
- 391
ReasoningOpenAI o3 · 88
- 195
- 295
- 395
MathOpenAI o3 · 99
- 199
- 299
- 399
CodingOpenAI o3 · 95
- 196
- 296
- 396
AgenticOpenAI o3 · 37
- 1100
- 2100
- 3100
Long contextOpenAI o3 · 73
- 183
- 283
- 381
Tool useOpenAI o3 · 78
- 189
- 287
- 387
Safety
- 192
- 292
- 391
Human preference
- 195
- 292
- 391
Newest receipts
13 recentLast 120 days
Benchmark rows added to the public ledger for OpenAI o3 in the last 120 days. Older rows live in the full table below.
| Benchmark | Family | Score | Source | Days ago |
|---|---|---|---|---|
| AA-LCR | Long context | 73.33% | Artificial Analysis | 5 |
| AA output speed | Performance | 134 t/s | Artificial Analysis | 5 |
| AA time to first token | Performance | 5.65s | Artificial Analysis | 5 |
| AIME | Math | 88.33% | Artificial Analysis | 5 |
| Artificial Analysis Intelligence Index | Reasoning | 31.1% | Artificial Analysis | 5 |
| GPQA Diamond | Reasoning | 82.7% | Artificial Analysis | 5 |
| Humanity's Last Exam | Reasoning | 20.1% | Artificial Analysis | 5 |
| IFBench | Reasoning | 71.43% | Artificial Analysis | 5 |
| LiveCodeBench | Coding | 80.8% | Artificial Analysis | 5 |
| MATH | Math | 99.2% | Artificial Analysis | 5 |
| MMLU-Pro | Knowledge | 85.3% | Artificial Analysis | 5 |
| SciCode | Coding | 41% | Artificial Analysis | 5 |
| Terminal-Bench | Agentic | 37.12% | Artificial Analysis | 5 |
Family coverage
Best score per family
A high composite that hides a weak family is a trap. These bars surface the families where this model hasn't been publicly tested, and where it leads.
Price vs. performance
Where OpenAI o3 sits on the scatter
Hover any dot for name, score, and input price. Keyboard: tab through the top twelve, or use the ranking below.
AnthropicOpenAIMistralAlibabaDeepSeekGoogleMoonshotMetaMiniMaxxAIZhipuXiaomiOtherMeituanCohereNVIDIA
Full benchmark ledger
Every catalog benchmark for OpenAI o3. Scores link to the original source; gaps mean no public row exists yet.
18 of 37 catalog benchmarks have a sourced row for OpenAI o3.
Knowledge
Reasoning
ARC-AGIVerified
87.5%
ARC-AGI-1 high compute
31.1%
AA Intelligence Index v4.1
GPQA DiamondVerified
82.7%
GPQA Diamond (AA run)
Humanity's Last ExamVerified
20.1%
HLE (AA run)
IFBenchVerified
71.43%
IFBench (AA run)
68.5%
LiveBench
Math
Coding
94.8% pass@1
HumanEval
LiveCodeBenchVerified
80.8%
LiveCodeBench (AA run)
SciCodeVerified
41%
SciCode (AA run)
Agentic
Terminal-BenchVerified
37.12%
Terminal-Bench Hard (AA run)
Long context
AA-LCRVerified
73.33%
AA-LCR (AA run)
51.2%
LongBench v2
Tool use
78.1%
BFCL
Performance
AA output speedVerified
134 t/s
Median output tokens/s (1k prompt, default provider)
AA time to first tokenVerified
5.65s
Median time to first token (1k prompt, default provider)
Sources
Where the numbers come from
- OpenAI o3 announcement· checked 2026-06-09
- LiveBench leaderboard· checked 2026-06-09