GPQA DiamondVerified
82.7%
GPQA Diamond (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.
Reasoning-tier OpenAI pick when latency and cost are secondary to hardest single-shot problem solving.
Composite53
Context200K
Input / 1M—
Output / 1M—
Knowledge cutoff2024-06-01
Output speed129 t/s
Statuslightweight
Family profile
Best published score in each covered benchmark family.
65/100 avg
- Reasoning
- 83
- Agentic
- 37
- Long context
- 75
3 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 |
|---|---|---|---|---|---|
| AA output speed | Performance | #40/ 195 | Artificial Analysis | 2026-10-12 | |
| AA-LCR | Long context | #90/ 184 | 75 | Artificial Analysis | 2026-09-28 |
| Terminal-Bench | Agentic | #111/ 167 | 37 | Artificial Analysis | 2026-09-28 |
| GPQA Diamond | Reasoning | #125/ 184 | 83 | Artificial Analysis | 2026-09-28 |
| Humanity's Last Exam | Reasoning | #128/ 186 | 20 | Artificial Analysis | 2026-09-28 |
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.
ReasoningOpenAI o3 · 83
- 196
- 295
- 395
Math
- 152
- 252
- 350
AgenticOpenAI o3 · 37
- 191
- 289
- 389
Long contextOpenAI o3 · 75
- 189
- 286
- 385
Newest receipts
5 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 output speed | Performance | 129 t/s | Artificial Analysis | 10 |
| AA-LCR | Long context | 74.67% | Artificial Analysis | 24 |
| GPQA Diamond | Reasoning | 82.7% | Artificial Analysis | 24 |
| Humanity's Last Exam | Reasoning | 20.1% | Artificial Analysis | 24 |
| Terminal-Bench | Agentic | 37.12% | Artificial Analysis | 24 |
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.
AnthropicOpenAIGoogleAlibabaMetaZhipuxAIDeepSeekMiniMaxXiaomiMoonshotOtherMeituanMistralNVIDIACohere
Full benchmark ledger
Every catalog benchmark for OpenAI o3. Scores link to the original source; gaps mean no public row exists yet.
5 of 13 catalog benchmarks have a sourced row for OpenAI o3.
Reasoning
Humanity's Last ExamVerified
20.1%
HLE (AA run)
Agentic
Terminal-BenchVerified
37.12%
Terminal-Bench Hard (AA run)
Long context
AA-LCRVerified
74.67%
AA-LCR (AA run)
Performance
AA output speedVerified
129 t/s
Median output tokens/s (1k prompt, default provider)
Sources
Where the numbers come from
- OpenAI o3 announcement· checked 2026-06-09
- LiveBench leaderboard· checked 2026-06-09
Questions this model page answers
Every answer below is assembled from the dated fields on this page. Nothing is written separately for search.
What is OpenAI o3?
OpenAI o3 is a OpenAI model: OpenAI's April 2025 reasoning model. Extended thinking for math, code, and science.
What is OpenAI o3 for?
Reasoning-tier OpenAI pick when latency and cost are secondary to hardest single-shot problem solving.
What does OpenAI o3 cost?
OpenAI o3 lists — input and — output per 1M tokens. Pricing last checked 2026-06-09.
How large is the OpenAI o3 context window?
OpenAI o3 publishes a 200K token context window.
What is the VerdictPal status of OpenAI o3?
OpenAI o3 carries Lightweight status on the model atlas.
The Pack · Editorial newsletter
Know when the rules move.
One short email when a university changes its AI rule for theses, or a research tool changes its pricing or privacy terms. No tracking, no third-party analytics. Unsubscribe in one click.