EN
Overview
OpenAI introduces GPT-5, its best AI system yet. GPT-5 is a significant leap in intelligence over all previous models, featuring state-of-the-art performance across coding, math, writing, health, visual perception, and more.
Benchmark Performance
GPT-5 sets new state-of-the-art results across multiple benchmarks:
- Math (AIME 2025, no tools): 94.6%
- Real-world coding (SWE-bench Verified): 74.9%
- Coding (Aider Polyglot): 88%
- Multimodal understanding (MMMU): 84.2%
- Health (HealthBench Hard): 46.2%
Key Capabilities
GPT-5 produces high-quality code, generates front-end UI with minimal prompting, and shows improvements to personality, steerability, and executing long chains of tool calls.
The model also introduces "minimal" reasoning and a "verbosity" parameter, giving users more control over how the model thinks and responds.
Model Family
The GPT-5 launch, combined with the previously released GPT-5.5 series, forms the complete GPT-5 model family:
- GPT-5.5 & GPT-5.5 Pro: Available in the API as of April 24, 2026, excelling at writing and debugging code, researching online, analyzing data, creating documents and spreadsheets, operating software.
- GPT-5.2: Focused on science and math, achieving 92.4% on GPQA Diamond and 40.3% on FrontierMath (Tier 1-3).
- GPT-5: The flagship model with broadest capabilities.
中文
概述
OpenAI 发布 GPT-5,称之为迄今为止最智能的 AI 系统。GPT-5 在编程、数学、写作、健康、视觉感知等多个领域都达到了最先进水平,相比此前所有模型都有显著智能飞跃。
基准测试表现
GPT-5 在多项基准测试中创下最新最佳成绩:
- 数学(AIME 2025,不使用工具): 94.6%
- 真实世界编程(SWE-bench Verified): 74.9%
- 编程(Aider Polyglot): 88%
- 多模态理解(MMMU): 84.2%
- 健康(HealthBench Hard): 46.2%
核心能力
GPT-5 能生成高质量代码,只需少量提示就能生成前端 UI,并在个性、可控性以及执行长链工具调用方面都有改进。
该模型还引入了"最小化"推理模式和" verbosity"参数,让用户能更好地控制模型的思考方式和回应方式。
模型家族
GPT-5 与此前发布的 GPT-5.5 系列共同构成完整的 GPT-5 模型家族:
- GPT-5.5 & GPT-5.5 Pro: 2026 年 4 月 24 日在 API 中上线,擅长编写和调试代码、在线研究、数据分析、创建文档和电子表格、操作软件。
- GPT-5.2: 专注于科学和数学,在 GPQA Diamond 上达到 92.4%,在 FrontierMath(Tier 1-3)上达到 40.3%。
- GPT-5: 旗舰模型,拥有最广泛的能力。