o3
Parameters
Undisclosed
Context Length
128K
Modality
Text
Architecture
Undisclosed
License
Proprietary
Release Date
20 Dec 2025
Knowledge Cutoff
Jun 2024
API Pricing (per 1M)
Input: $2.00 · Output: $8.00
Evaluation Benchmarks
Rank
#69
| Benchmark | Score | Rank |
|---|---|---|
Software Engineering | 0.58 | 28 |
Graduate-Level QA | 0.833 | 51 |
General Text | 1431 | 73 |
Intelligence Index | 0.20 | 138 |
QA Assistant Archived | 0.986 | 🥇 1 |
Coding Archived | high 0.81 Standard 0.77 | 🥈 2 4 |
Rankings
Overall Rank
#69
Coding Rank
#71
About o3
o3 provides advanced reasoning capabilities for complex problem-solving across multiple domains. Features deliberative thinking for mathematics, coding, and analytical tasks. Achieves strong performance on challenging benchmarks including competitive programming, advanced mathematics, and scientific reasoning. Well-suited for applications requiring careful analysis and multi-step reasoning at a balanced cost-performance ratio.
Technical Specifications
Architecture specifications are undisclosed for proprietary models.
Attention
Attention Structure
Multi-Head Attention
Attention Heads
Undisclosed
Key-Value Heads
Undisclosed
Attention Head Dimension
Undisclosed
Position Embedding
Absolute Position Embedding
RoPE Theta
Undisclosed
Sliding Window Attention
Undisclosed
Sliding Window Size
Undisclosed
Sliding Window Ratio
Undisclosed
Linear Attention
Undisclosed
Linear Attention Ratio
Undisclosed
Normalization
Undisclosed
Activation Function
Undisclosed
Dimensions
Hidden Dimension Size
Undisclosed
Number of Layers
Undisclosed
FFN Intermediate Size (Dense)
Undisclosed
Multi-Token Prediction Heads
Undisclosed
Tokenizer
Vocabulary Size
Undisclosed
Resources
About o3
OpenAI's o3 reasoning models represent a breakthrough in deliberative problem-solving and mathematical reasoning. These models use advanced chain-of-thought techniques and can be configured with different compute levels (low, medium, high) to balance reasoning depth with response time. Excel at complex mathematics, scientific reasoning, and multi-step problem solving.
Other o3 Models
- No related models available