OpenAI o3
FrontierMultimodalAPI Available
Benchmark Scores
95.1/100MMLU93.8
HumanEval94.5
Math96.2
Reasoning96.8
Coding94
About OpenAI o3
OpenAI o3 is the full-scale reasoning model succeeding o1, delivering breakthrough performance on ARC-AGI, competition math, and doctoral-level science. It uses advanced chain-of-thought with configurable compute budgets.
Strengths
- +Top-tier mathematical reasoning
- +ARC-AGI breakthrough performance
- +Strong at competition programming
- +Excellent scientific reasoning
Weaknesses
- -Very high latency on hard problems
- -Expensive for casual use
- -Overthinks simple queries
Pricing
Per 1M tokens
Input$10.00
Output$40.00
Quick Facts
- Context Window
- 200K
- Parameters
- Undisclosed
- Release Date
- 2025-07-15
- Category
- Frontier
Compare Models
See how OpenAI o3 stacks up against other models