1 million business customers putting AI to work

今天我们宣布，全球已有超过 1,000,000 家企业客户直接在使用 OpenAI —— 这是有史以来增长最快的企业平台。这里统计的是所有为商业用途主动付费使用 OpenAI 的组织——无论是通过 ChatGPT for Work，还是通过我们的开发者平台直接调用模型。我们以能与金融服务、医疗、零售等行业的领军企业合作为荣；在这些行业中，我们的技术正把智能能力嵌入到客户体验、内部运营和团队工作流程的核心位置。

从个人使用到公司级影响

企业端的势头部分来自于消费者端的普及。现已有超过 8 亿每周用户熟悉 ChatGPT，这让企业内部的采用和投资回报更快显现——试点周期更短、部署阻力更小。我们在 ChatGPT for Work 的增长上已经看到这一点：目前 ChatGPT for Work 的席位总数超过 7,000,000 个，仅两个月内就增长了 40% 。而 ChatGPT Enterprise 的席位数同比增长了 9 倍。

面向企业的新工具、集成和能力

为支持这种加速，我们推出了一系列新工具、集成和能力，帮助团队从早期试用快速迈向大规模部署：

• 我们推出了 Company Knowledge ，让 ChatGPT 能跨越 Slack、SharePoint、Google Drive、GitHub 等工具进行推理、回答问题、做分析并采取行动，所用的是为工具协作和提供引用优化的 GPT‑5 版本。
• 面向代码生成、重构和工作流自动化的 Codex 的采用速度也大幅提速——自 8 月以来使用量增长了 10 倍。像 Cisco 这样的组织已将 Codex 纳入工程工作流，代码审查时间缩短了 50%，项目周期从数周压缩到数日。
• 我们通过 AgentKit 让企业级代理的构建与部署变得可行，使团队能够在数天而非数月内将想法推向生产。像 Carlyle 使用的 AgentKit 评估平台，将其多代理尽职调查框架的开发时间缩短了超过 50%，并将代理的准确率提高了 30%。
• 我们还推进了多模态模型，支持更丰富的工作流——从用于视觉和视频创作的 Image Generation API 和 Sora 2，到用于构建生产级语音代理的 gpt‑realtime 与 Realtime API。各行各业的团队现在可以在同一系统中处理文本、图像、视频和音频。

企业在获得实质性回报

人工智能正在带来真实的商业价值。根据最近一项来自 Wharton 的研究，75% 的企业报告获得了正向 ROI，而报告负回报的不到 5%。这一结果也反映了我们在客户中看到的实际情况：当 AI 与合适的用例和基础设施结合时，团队会看到切实成效。

• Indeed 在其 Invite to Apply 功能中使用 OpenAI API，使申请量提高了 20%，录用提升了 13%。
• Lowe’s 在超过 1,700 家门店中通过 Mylow Companion（基于 OpenAI 模型的店内应用）为所有员工提供专业项目指导。
• 将 OpenAI 作为骨干的 Fin，其客户服务代理 Intercom 的开发周期已从数季度缩短为数日。
• 从今天起， Databricks 正把 OpenAI 的前沿智能带到企业数据所在之处——让构建和运行高质量代理变得更容易。

我们的下一个方向

除了内部采用外，我们还看到第二条趋势：越来越多的企业希望在 OpenAI 平台上直接构建。

公司正在我们的平台上创建新的应用和具代理性的工作流。 Canva、 Figma、 Zillow、 Spotify 等已将它们的应用直接接入 ChatGPT，让用户在惯常的环境中获得服务。 Shopify、 Etsy、 Walmart、 PayPal 和 Salesforce 正通过在 ChatGPT 中采用 Agentic Commerce Protocol（ACP）构建新的购物体验，把对话式商务融入日常决策流程。

我们由衷感谢今天与我们一起构建的客户，并期待 2026 年及以后的发展。重新思考工作操作系统是一个巨大机遇，我们也很期待打造支撑这一未来的平台。

Today, we’re announcing that more than 1 million business customers around the world are directly using OpenAI—the fastest-growing business platform in history. This includes all organizations that actively pay OpenAI for business use—either through ChatGPT for Work, or through direct consumption of our models through our developer platform. We’re proud to work with category leaders in industries like financial services, healthcare, retail, and more, where our technology is making intelligence central to their customer experiences, internal operations, and team-level workflows.

From personal use to companywide impact

Our enterprise momentum is fueled in part by consumer adoption. With more than 800 million weekly users already familiar with ChatGPT, adoption and ROI within businesses is realized more rapidly—pilots are shorter and rollouts face less friction. We’re seeing this with the growth of ChatGPT for Work—we now have more than 7 million total ChatGPT for Work seats (up 40% in just 2 months⁠). ChatGPT Enterprise seats specifically have grown 9x year-over-year.

New tools, integrations, and capabilities for businesses

To support this acceleration, we’ve launched a new wave of tools, integrations, and capabilities to help teams move from early use to full-scale deployment:

• We introduced company knowledge⁠, which lets ChatGPT reason across tools like Slack, SharePoint, Google Drive, GitHub, and more to get answers, do analysis, and take action using a version of GPT‑5 that’s optimized for working with tools and providing citations.
• Adoption for Codex⁠, our model for code generation, refactoring, and workflow automation has accelerated quickly—usage is up 10x since August. Recently, organizations like Cisco have rolled out Codex into their engineering workflows, cutting code review times by 50% and shrinking project timelines from weeks to days.
• We also made enterprise agents practical to build and deploy with AgentKit⁠, making it possible for teams to go from idea to production in days instead of months. Companies like Carlyle are already seeing results: the AgentKit evaluation platform cut development time on Carlyle’s multi-agent due diligence framework by over 50% and improved agent accuracy by 30%.
• We also advanced our multimodal models to enable richer workflows—from the Image Generation API⁠ and Sora 2⁠ for visual and video creation to gpt-realtime and Realtime API⁠ to build production voice agents. Teams across every industry can now work across text, images, video, and audio in one system.

Companies are seeing meaningful ROI  

AI is delivering real business value. ⁠According to a recent Wharton study⁠, 75% of enterprises report a positive ROI, and fewer than 5% report a negative return. While there are many studies on this topic, this one reflects what we see on the ground today with our customers: when AI is deployed with the right use case and infrastructure, teams see real results.

• Indeed⁠ is using OpenAI APIs in its Invite to Apply feature to drive a 20% increase in applications and a 13% lift in hires. 
• Lowe’s⁠ empowers all associates in 1,700+ stores with expert project guidance through Mylow Companion, an in-store app built with OpenAI models.
• With OpenAI as the backbone of Fin, their customer service agent, Intercom⁠ has accelerated development cycles from quarters to days.  
• And starting today, Databricks⁠ is bringing OpenAI frontier intelligence to where enterprises’ data already lives—making it easier to build and run high-quality agents.

Where we’re heading next

Beyond internal adoption, we’re seeing a second trend: businesses increasingly want to build with OpenAI too.

Companies are creating new applications and agentic workflows directly on our platform. Canva, Figma, Zillow, Spotify, and others have plugged their apps⁠ directly into ChatGPT to meet users where they already are. Shopify, Etsy, Walmart, PayPal, and Salesforce are building new shopping experiences through the Agentic Commerce Protocol (ACP)⁠ in ChatGPT, bringing conversational commerce into the flow of everyday decisions.

We’re deeply grateful to the customers building with us today—and we’re looking forward to what’s next in 2026 and beyond. There’s a big opportunity to rethink the operating system for work, and we’re excited to build the platform to power it.

Generated by RSStT. The copyright belongs to the original author.

Source