ChatGPT Work
已验证OpenAI's agent inside ChatGPT that gathers context across your apps and ships finished work
发布时间
2026
最近更新
2026-08-31
国家/地区
United States
API
不可用
自托管
否
融资
Venture-backed (OpenAI)
最近更新
2026-08
关于 ChatGPT Work
总评
The default general-purpose agent for the ChatGPT installed base. Not the deepest integration surface, but the lowest-friction path from goal to finished deliverable — and the GPT-5.6 Sol engine makes it genuinely capable rather than a demo.
功能特性
详细评分
优缺点
优点
- Zero new tooling — lives inside ChatGPT where hundreds of millions already work
- Ships finished deliverables (reports, decks, prototypes), not just answers
- Backed by OpenAI's frontier GPT-5.6 Sol model
缺点
- Capability ceiling set by ChatGPT's connector ecosystem, not by you
- Gating and usage limits vary by plan and are not fully documented
- Strict data-boundary or self-hosted workflows need Enterprise
应用场景
适合谁使用?
Knowledge workers already in the ChatGPT ecosystem who want outcomes delegated, not just answers
常见问题
What is ChatGPT Work?
ChatGPT Work is OpenAI's agent mode inside ChatGPT, launched July 2026. It gathers context across your connected apps, breaks a goal into steps, and ships finished work. It runs on the GPT-5.6 Sol model.
How much does ChatGPT Work cost?
It is included with existing ChatGPT tiers rather than sold separately — availability and usage limits scale with your plan (Free through Enterprise). Check OpenAI's plan page for current gating.
How is it different from Claude or other agents?
ChatGPT Work optimizes for ubiquity: no new tool, connector-based app context, deliverable-focused output. Dedicated agent platforms offer deeper integrations and data control, at the cost of setup effort.
相关智能体
Runable
效率智能体
General agent that builds websites, apps, and decks from a prompt — then keeps operating them
Bardeen
效率智能体
面向知识工作者的 AI 工作流自动化
Career Ops
效率智能体
开源 AI 驱动的求职系统
Claude in Chrome
效率智能体
Anthropic's browser agent that navigates, fills forms, and completes tasks inside Chrome
热门替代方案
与这些相似工具进行比较
链接与资源
档案由 AI 辅助、基于公开信息生成——未经独立手动测试。