OpenAI o3-mini
OpenAI o3-mini is an AI model published by OpenAI, released on 2025-01-31, for 推理大模型, with 0.0B parameters, and 200K tokens context length, under the 不开源 license.
Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology
| Modality | Input | Output |
|---|---|---|
| Text | $1.1 | $4.4 |
OpenAI o3-mini currently shows benchmark results led by AIME2025 (47 / 107, score 86.50), MATH-500 (24 / 43, score 95.80), GPQA Diamond (99 / 166, score 70.60). This page also consolidates core specs, context limits, and API pricing so you can evaluate the model from benchmark results and deployment constraints together.
OpenAI o3 mini是OpenAI发布的一个OpenAI o3模型的较小参数规模版本,速度更快。
OpenAI o3 mini是OpenAI发布的第二代推理大模型中最小规模的版本。第一代推理大模型是o1,跳过了o2版本号的原因可能是o2与英国电信运营商的品牌冲突。o3 mini总共分为三个版本,o3-mini low,o3-mini medium 和o3-mini high。本模型为其中最强的版本,即o3-mini high。ChatGPT中免费用户使用的是o3-mini meidum版本。Plus用户可以选择这个版本。同时,根据测试,这个版本的模型也比之前的o1更强。
不同版本的评测结果如下:
| Category | 01-mini | 03-mini (low) | 03-mini (medium) | 03-mini (high) |
|---|---|---|---|---|
| General | 85.2 | 84.9 | 85.9 | 86.9 |
| Math | 90.0 | 95.8 | 97.3 | 97.9 |
| Factuality | 89.9 | 55.1 | 90.8 | 92.0 |
| Eval | 76.0 | 13.0 | 13.4 | 13.8 |
| Model | Reasoning Level | Average | LCB Generation | Code Completion |
| o3-mini | low | 0.618 | 0.756 | 0.48 |
| o3-mini | medium | 0.723 | 0.846 | 0.60 |
| o3-mini | high | 0.846 | 0.820 | 0.833 |
| o1-mini | high | 0.674 | 0.628 | 0.72 |
欢迎关注 DataLearner 官方微信,获得最新 AI 技术推送
