DataLearner logoDataLearnerAI
Latest AI Insights
Model Leaderboards
Benchmarks
Model Directory
Model Comparison
Resource Center
Tools
LanguageEnglish
DataLearner logoDataLearner AI

A knowledge platform focused on LLM benchmarking, datasets, and practical instruction with continuously updated capability maps.

Products

  • Leaderboards
  • Model comparison
  • Datasets

Resources

  • Tutorials
  • Editorial
  • Tool directory

Company

  • About
  • Privacy policy
  • Data methodology
  • Contact

© 2026 DataLearner AI. DataLearner curates industry data and case studies so researchers, enterprises, and developers can rely on trustworthy intelligence.

Privacy policyTerms of service
Page navigation
目录
Model catalogOpenAI o4 - mini
OP

OpenAI o4 - mini

推理大模型

OpenAI o4 - mini

Release date: 2025-04-16更新于: 2025-04-19 10:57:281,263
Live demoGitHubHugging FaceCompare
Parameters
Not disclosed
Context length
200K
Chinese support
Supported
Reasoning ability

OpenAI o4 - mini is an AI model published by OpenAI, released on 2025-04-16, for 推理大模型, 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

OpenAI o4 - mini

Model basics

Reasoning traces
Supported
Thinking modes
Thinking modes not supported
Context length
200K tokens
Max output length
No data
Model type
推理大模型
Release date
2025-04-16
Model file size
No data
MoE architecture
No
Total params / Active params
No data / N/A
Knowledge cutoff
No data
OpenAI o4 - mini

Open source & experience

Code license
不开源
Weights license
不开源- 不开源
GitHub repo
GitHub link unavailable
Hugging Face
Hugging Face link unavailable
Live demo
No live demo
OpenAI o4 - mini

Official resources

Paper
Introducing OpenAI o3 and o4-mini
DataLearnerAI blog
No blog post yet
OpenAI o4 - mini

API details

API speed
3/5
💡Default unit: $/1M tokens. If vendors use other units, follow their published pricing.
Standard pricingStandard
ModalityInputOutput
Text$1.1$4.4
Image$1.1--
OpenAI o4 - mini

Benchmark Results

OpenAI o4 - mini currently shows benchmark results led by AIME 2024 (1 / 62, score 98.70), MMLU (2 / 65, score 93), AIME2025 (10 / 106, score 99.50). This page also consolidates core specs, context limits, and API pricing so you can evaluate the model from benchmark results and deployment constraints together.

Thinking
All modesThinking
Tool usage
All modesWith toolsNo tools

综合评估

6 evaluations
Benchmark / mode
Score
Rank/total
MMLU
Thinking Mode
93
2 / 65
GPQA Diamond
Thinking Mode
81.40
63 / 175
MMLU Pro
Thinking Mode
80.60
53 / 124
ARC-AGI
Thinking Mode
58.70
27 / 56
HLE
Thinking Mode
14.28
114 / 148
HLE
Thinking ModeTools
17.70
102 / 148

编程与软件工程

2 evaluations
Benchmark / mode
Score
Rank/total
CodeForces
Thinking ModeTools
2719
6 / 16
SWE-bench Verified
Thinking Mode
68.10
59 / 103

数学推理

12 evaluations
Benchmark / mode
Score
Rank/total
AIME2025
Thinking Mode
92.70
32 / 106
AIME2025
Thinking ModeTools
99.50
10 / 106
AIME 2024
Thinking Mode
93.40
5 / 62
AIME 2024
Thinking ModeTools
98.70
1 / 62
FrontierMath
Low
9.70
26 / 57
FrontierMath
Medium
19.30
15 / 57
FrontierMath
High
17.20
18 / 57
IMO-ProofBench
High
11.40
12 / 16
IMO 2024
Thinking Mode
7.70
7 / 10
FrontierMath - Tier 4
Medium
2.10
24 / 38
FrontierMath - Tier 4
High
6.30
15 / 38
IMO 2025
Thinking Mode
3
7 / 9

常识推理

1 evaluations
Benchmark / mode
Score
Rank/total
Simple Bench
Thinking Mode
38.70
19 / 27

Agent能力评测

3 evaluations
Benchmark / mode
Score
Rank/total
Aider-Polyglot
High
72
8 / 26
τ²-Bench
Thinking ModeTools
56.90
30 / 40
τ²-Bench - Telecom
Thinking ModeTools
50.20
33 / 35
View benchmark analysisCompare with other models
OpenAI o4 - mini

Publisher

OpenAI
OpenAI
View publisher details
OpenAI o4 - mini

Model Overview

o4 mini是OpenAI最新发布的推理大模型。

OpenAI o4-mini 是一款专注于快速、经济高效推理的小型化模型。尽管其规模较小,但它在数学、编码和视觉任务等领域展现出显著的性能。

该模型具备强大的推理能力,并能够有效地利用和组合ChatGPT内的各种工具,包括网络搜索、使用Python分析上传文件和数据、对视觉输入进行深度推理,甚至生成图像。o4-mini经过训练,能够判断何时以及如何使用这些工具来解决复杂问题,并生成详细且经过深思熟虑的答案。

在性能方面,o4-mini在多个基准测试中表现出色。例如,在AIME 2024和2025数学竞赛中,o4-mini是表现最佳的基准模型。当配合Python解释器使用时,o4-mini在AIME 2025上实现了99.5%的pass@1(8个一致性答案下达到100%的共识)。这体现了其有效利用工具的能力。专家评估指出,o4-mini不仅在数学、编码和视觉任务上表现优异,在非STEM任务以及数据科学等领域也超越了其前代模型o3-mini。同时,与前代推理模型相比,o4-mini在指令遵循、提供更有用和可验证的回复方面均有提升,交互时也表现得更为自然和对话化,能够更好地利用记忆和过往对话内容使回复更具个性化和相关性。

o4-mini在效率和成本方面也具有优势。由于其高效性,o4-mini支持比o3更高的使用限制,使其成为处理需要推理能力的高容量、高吞吐量任务的有力选择。在成本效益方面,o4-mini相较于o3-mini实现了提升,预计在多数实际应用场景中,o4-mini将比o3-mini更智能且更经济。

在安全性方面,OpenAI为o3和o4-mini重建了安全训练数据,增加了在生物风险、恶意软件生成和越狱等领域的拒绝提示。这使得o4-mini在内部拒绝基准测试中表现出色。同时,OpenAI还开发了系统级缓解措施来标记高风险领域的危险提示。根据评估结果,o4-mini在生物与化学、网络安全和AI自我改进三个追踪能力领域均低于“高”风险阈值。

用户可以通过多种途径访问o4-mini。ChatGPT Plus、Pro和Team用户可以在模型选择器中找到o4-mini和o4-mini-high,它们将替代此前的o3-mini和o3-mini-high。免费用户可以在提交查询前选择“Think”来体验o4-mini。开发者也可以通过Chat Completions API和Responses API使用o4-mini。

DataLearner 官方微信

欢迎关注 DataLearner 官方微信,获得最新 AI 技术推送

DataLearner 官方微信二维码