GPT-5.3 and GPT-5.5 in ChatGPT

GPT-5.3 and GPT-5.5 in ChatGPT

Summary (EN)

OpenAI says GPT-5.3 is now the default ChatGPT model for all logged-in users, while GPT-5.5 and GPT-5.5 Pro are rolling out to paid tiers with a more explicit split between fast, auto-switching usage and deeper reasoning modes. The help article describes GPT-5.3 Instant as the standard experience for everyday work, including information seeking, technical writing, translation, and general assistance. For harder tasks, ChatGPT can automatically route requests from Instant to GPT-5.5 Thinking, which OpenAI positions as better at complex goal understanding, tool use, multi-step task completion, spreadsheet work, document understanding, research across web sources, image understanding, and cleaner instruction following. GPT-5.5 Pro is presented as the highest-capability option for long-running or especially difficult workflows, though it does not currently support every ChatGPT feature. The article also explains a new operating model around manual model selection, auto-switching behavior, thinking traces, context windows, plan-based access, and message limits across Free, Go, Plus, Business, Pro, Enterprise, and Edu tiers. In practical terms, this is both a product release and a packaging shift. OpenAI is consolidating the default ChatGPT experience around one front-door model while reserving premium reasoning depth, longer context, and more configurable thinking effort for higher-paid users. That makes the release directly relevant to application builders evaluating when deeper reasoning should be surfaced as a premium feature versus silently invoked behind a simpler default interface.

Summary (ZH)

OpenAI 表示，GPT-5.3 已成为所有已登录 ChatGPT 用户的默认模型，而 GPT-5.5 与 GPT-5.5 Pro 正在向付费层级逐步推出，形成“默认快速体验”和“更深度推理模式”之间更明确的产品分层。帮助文档将 GPT-5.3 Instant 定位为日常使用的标准入口，适合信息检索、技术写作、翻译和通用助手场景。对于更复杂的问题，ChatGPT 可以自动把请求从 Instant 路由到 GPT-5.5 Thinking。OpenAI 将后者描述为更擅长理解复杂目标、调用工具、完成多步骤任务，以及处理表格、文档、跨网页研究、图像理解和更严格的指令遵循。GPT-5.5 Pro 则被定义为面向长流程、高难度工作的最高能力选项，但当前并不支持 ChatGPT 的全部功能。文章同时详细说明了新的模型选择方式、自动切换逻辑、推理轨迹展示、上下文窗口，以及 Free、Go、Plus、Business、Pro、Enterprise、Edu 等不同套餐下的配额和权限差异。从产品意义上看，这不仅是一次模型更新，也是一次交互和商业打包的调整。OpenAI 正把普通用户入口统一到一个默认前台模型，同时将更深度的推理能力、更长上下文和更细粒度的思考控制，作为高阶付费能力提供。这对应用侧很重要，因为它意味着“何时显式暴露高级推理、何时在后台自动启用”正在成为商业化产品设计的一部分。

Source

https://help.openai.com/en/articles/11909943-gpt-53-and-gpt-55-in-chatgpt