OpenAI发射GPT-5.4,具有先进的推理、编码和自主任务执行能力,现住在ChatGPT和APIs。 OpenAI launches GPT-5.4 with advanced reasoning, coding, and autonomous task execution, now live in ChatGPT and APIs.

OpenAI已发布GPT-5.4, 这是一项重大升级, 包括100万吨的上下文窗口、强化的电脑视觉及本地桌面控制, OpenAI has released GPT-5.4, a major upgrade with a 1-million-token context window, enhanced computer vision, and native desktop control enabling autonomous task execution across apps. 与GPT-5.2相比,它减少了33%的错误,在OSWorld-Verificed和APEX-Agencys等基准方面成绩优异,并引入实时调整的工具发现和思考模式。 It reduces errors by 33% compared to GPT-5.2, excels in benchmarks like OSWorld-Verified and APEX-Agents, and introduces tool discovery and Thinking mode for real-time adjustments. 它提供标准版本、Pro版本和思考版本,支持在编码、财务和研究方面的高级工作流程,提高效率和降低成本。 Available in Standard, Pro, and Thinking versions, it supports advanced workflows in coding, finance, and research, with improved efficiency and lower costs. 该模型正在整个ChatGPT、Codex和API推出,为企业和开发商定价。 The model is rolling out across ChatGPT, Codex, and the API, with pricing tiers for enterprise and developers.