o3 Pro
OpenAI · released Jun 10, 2025
The o-series of models are trained with reinforcement learning to think before they answer and perform complex reasoning. The o3-pro model uses more compute to think harder and provide consistently...
Specification
- Context window
- 200K
- Max output
- 100K
- Knowledge cutoff
- Jun 30, 2024
- Parameters
- Undisclosed
- Licence
- Proprietary
- Serving providers
- 1
- Moderated
- Yes
Intelligence
21.9
70th percentile
Coding
—
Coding Index
Agentic
—
Agentic Index
Output speed
32 t/s
Median across providers
Latency
23.39s
Time to first token
Cost per task
$7.40
Estimated
Benchmarks
Where the score comes from
The Intelligence Index is a composite. These are the underlying evaluations this model was actually measured on.
Evaluation scores
Percentage correct · higher is better
- GPQA Diamond84.5%
An evaluation missing from this list was not run for this model — it is not a zero.
View as table
| Evaluation | Score |
|---|---|
| GPQA Diamond | 84.5% |
Against its peers
Intelligence Index · this model highlighted, nearest peers in grey
- M2.7MiniMax22.8
- Hy3 previewTencent22.7
- Gemini 3.5 Flash LiteGoogle22.2
- GLM 4.7Z.ai22.2
- Kimi K2 ThinkingMoonshot AI22.0
- o3 ProOpenAI21.9
- MiMo-V2.5Xiaomi21.8
- Qwen3.6 27BQwen21.4
Peers are the models sitting closest on the Intelligence Index — the set you would realistically choose between.
View as table
| Model | Intelligence |
|---|---|
| M2.7 | 22.8 |
| Hy3 preview | 22.7 |
| Gemini 3.5 Flash Lite | 22.2 |
| GLM 4.7 | 22.2 |
| Kimi K2 Thinking | 22.0 |
| o3 Pro | 21.9 |
| MiMo-V2.5 | 21.8 |
| Qwen3.6 27B | 21.4 |
Percentile among all indexed models
Pricing
What it costs to run
List prices per million tokens, plus what one representative task works out to.
List price
- Input / 1M tokens
- $20
- Output / 1M tokens
- $80
- Cached input / 1M
- Not offered
- Blended 3:1
- $35
One task, estimated
$7.40
- Input tokens
- 50,000
- Output tokens
- 80,000
- Profile
- Reasoning
Estimated from list pricing: 50K input tokens plus 80K output tokens for reasoning models (25K for non-reasoning).
Serving providers
Speed and latency figures are medians across these providers, so a widely-served model reports a blend rather than any single endpoint.