Codex is now generally available

我们宣布 Codex 的全面可用性，并推出三项新功能，使其对工程团队更有用：

• 全新的 Slack 集成：可以像对同事那样，在团队频道或线程中直接将任务委派给 Codex 或向其提问。
• Codex SDK：将为 Codex CLI 提供动力的同一代理嵌入到你的工作流、工具和应用中，对 GPT‑5‑Codex（https://openai.com/index/introducing-upgrades-to-codex/）等模型提供最先进的性能，无需额外微调。
• 新的管理员工具：通过环境控制、监控和分析仪表板，ChatGPT 工作区管理员现在在大规模管理 Codex 时拥有更多可见性和控制权。

自从 Codex 云代理（https://openai.com/index/introducing-codex/）在五月以研究预览形式上线以来，Codex 已稳步演进为更可靠、更有能力的编码协作者。现在你可以在任何编程场所与它协作——在编辑器、终端和云端，均通过你的 ChatGPT 账号连接。自八月初以来，Codex 的日常使用量增长了超过 10 倍，GPT‑5‑Codex 是我们增长最快的模型之一，上线后三周内已处理超过 40 万亿（40 trillion）个 token。

世界各地的开发者都在使用 Codex，从 Duolingo、Vanta 这样的初创公司到 Cisco、楽天（Rakuten）这样的企业。 在 OpenAI 内部，它已成为我们构建流程的核心：目前几乎所有工程师都在使用 Codex，而 7 月时仅略高于一半。工程师每周合并的 pull request 增加了 70%，Codex 自动审查几乎每个 PR，以在代码进入生产之前捕获关键问题。

Codex 的全面可用反映了开发者和团队对它的快速采用——以及它已经在改变我们构建方式方面的影响。

CODEX 在 SLACK 中

在 Slack 频道或线程中标注 @Codex，它会自动从对话中采集所需上下文，选择合适的环境，并返回带有指向 Codex 云中已完成任务链接的答复。随后你可以合并它的更改、继续迭代，或将任务拉到本地电脑继续开发。请在文档中了解如何设置 Slack 集成（在新窗口打开）：http://developers.openai.com/codex/integrations/slack

CODEX SDK

GPT‑5‑Codex 是专为 Codex 训练的——尤其是为为 Codex CLI 提供动力的开源代理实现。我们还对该代理实现进行了调优，使其提示、工具定义和代理循环在 GPT‑5‑Codex 等模型上提供更快、更准确的结果。

通过 Codex SDK，你只需几行代码就能将同一代理引入自己的工程工作流和应用。它提供用于解析代理响应的结构化输出和内置的上下文管理以恢复会话。该 SDK 今天已提供 TypeScript 版本，更多语言即将推出。

TypeScript 示例代码： import { Codex } from "@openai/codex-sdk";

const agent = new Codex({}); const thread = await agent.startThread();

const result = await thread.run("Explore this repo"); console.log(result);

// resume thread const result2 = await thread.run("Propose changes") console.log(result2);

我们还发布了一个新的 GitHub Action，便于将 Codex 集成到 CI/CD 管道中。如果你需要在 shell 环境中直接在工作流里使用 Codex 代理，也可以安装并通过 codex exec 运行 Codex CLI。请在我们的最新指南中了解有关使用 SDK 和 GitHub Actions 的更多信息（在新窗口打开）：http://developers.openai.com/codex/sdk

新的管理员功能

ChatGPT 管理员现在可以在其工作区编辑或删除 Codex 云环境——例如移除敏感信息或清理未使用的环境。管理员还可以对本地使用（Codex CLI 和 IDE 扩展）强制执行更安全的默认设置，例如通过托管配置定义覆盖或监控 Codex 所采取的操作。新的分析仪表板还可帮助管理员跟踪 CLI、IDE 和网页端的使用情况，以及 Codex 提供的代码审查质量。

在管理员指南中了解有关配置新控制项的更多信息（在新窗口打开）：https://developers.openai.com/codex/enterprise

（Codex 分析仪表板示意图：显示两张图表——按优先级分类的每日代码审查问题条形图，以及随时间变化的代码审查反馈情绪堆叠面积图。背景为渐变，带有淡淡的代码样式。）https://images.ctfassets.net/kftzwdyauwt9/2vBF5iAOIrcHXJeydbsMPW/1e76d314fd2cddc845272cc23deea5bc/Codex_Blog_UI-Dashboard__2_.png?w=3840&q=90&fm=webp

CODEX 的应用案例

CISCO：代码审查速度提升 50%

在 Cisco，工程师使用 Codex 来加速复杂 pull request 的审查，审查时间最多减少 50%。通过减少手动检查的时间，他们可以把更多精力投入到更有意义、具有变革性的工作上。Codex 既审查人工编写的代码，也审查其自身生成的输出，帮助工程师在实现更快、更大胆的产品发布的同时保持高标准。

INSTACART：自动化代码清理

在 Instacart，Codex SDK 与他们的后台编码代理平台 Olive 集成。工程师可以启动远程开发环境，并一键完成端到端任务，使用 Codex 编辑和测试更改。Codex 会自动清理技术负债，如死代码和过期的实验，提升代码质量并降低代码库的延迟。它还承担重复且明确的更改，减少积压工作并显著加速工程效率。

可用性和定价更新

Slack 集成和 Codex SDK 从今天起对 ChatGPT Plus、Pro、Business、Edu 和 Enterprise 计划的开发者开放，而新的管理员功能将对 Business、Edu 和 Enterprise 开放。

自 10 月 20 日起，Codex 云任务也将开始计入你的 Codex 使用量。了解各计划中 Codex 定价的更多信息（在新窗口打开）：https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan

We’re announcing the general availability of Codex and three new features that make it even more useful for engineering teams:

• A new Slack integration: Delegate tasks or ask questions to Codex directly from a team channel or thread, just like you would a coworker.
• Codex SDK: Embed the same agent that powers the Codex CLI into your own workflows, tools, and apps for state-of-the-art performance on GPT‑5-Codex⁠ without extra tuning.
• New admin tools: With environment controls, monitoring, and analytics dashboards, ChatGPT workspace admins now have more visibility and control to manage Codex at scale.

Since the Codex cloud agent⁠ launched in research preview in May, Codex has steadily evolved into a more reliable and capable coding collaborator. You can now work with it everywhere you code—in your editor, terminal, and the cloud, all connected by your ChatGPT account. Daily usage of Codex has grown by more than 10x since early August, and GPT‑5-Codex is one of our fastest growing models ever, serving over 40 trillion tokens in the three weeks since launch.

Codex is now used by developers all over the world, from startups like Duolingo and Vanta to large enterprises like Cisco and Rakuten. Inside OpenAI, it’s become integral to how we build: nearly all engineers use Codex today, up from just over half in July. They merge 70% more pull requests each week, and Codex automatically reviews almost every PR to catch critical issues before they reach production.

The general availability of Codex reflects how quickly developers and teams everywhere are adopting it—and how much it’s already changing the way we build. 

Codex in Slack

Tag @Codex in a Slack channel or thread and it will automatically gather the context it needs from the conversation, choose the right environment, and answer with a link to the completed task in Codex cloud. From there, you can merge its changes, keep iterating, or pull the task to your computer to keep working locally. Learn how to set up the Slack integration in the docs⁠. 

Codex SDK

GPT‑5-Codex was trained for Codex—specifically, the open-source agent implementation that powers the Codex CLI. We also tuned the agent implementation so that its prompt, tool definitions, and agent loop  deliver faster and more accurate results with models like GPT‑5-Codex.

With the Codex SDK, you can bring the same agent into your own engineering workflows and apps with just a few lines of code. It provides structured outputs for parsing agent responses and built-in context management to resume sessions. The SDK is available for TypeScript today, with more languages coming soon.

1import { Codex } from "@openai/codex-sdk";

2


3const agent = new Codex({});

4const thread = await agent.startThread();

5


6const result = await thread.run("Explore this repo");

7console.log(result);

8


9//resume thread

10const result2 = await thread.run("Propose changes")

11console.log(result2);

We’re also releasing a new GitHub Action to make it easy to integrate Codex into CI/CD pipelines. If you need to directly use the Codex agent in workflows that run in shell environments, you can just install and run the Codex CLI with codex exec. Learn more about using the SDK and GitHub Actions in our latest guide⁠.

New admin features

ChatGPT admins can now edit or delete Codex cloud environments in their workspace—for example, to remove sensitive information or clean up unused environments. They can also enforce safer defaults for local usage with the Codex CLI and IDE extension, like defining overrides through managed configuration or monitor actions taken by Codex. New analytics dashboards also help admins track usage across the CLI, IDE, and web, and the quality of code reviews provided by Codex.

Learn more about configuring the new controls in the admin guide⁠.

Codex at work

At Cisco, engineers are using Codex to speed up the review of complex pull requests, reducing review times by up to 50%. By spending less time on manual checks, they can devote more energy to meaningful, transformative work. Codex supports them by reviewing both human-written code and its own output, helping engineers maintain high standards while enabling faster, more ambitious product releases.

At Instacart, the Codex SDK is integrated with Olive, their background coding agent platform. Engineers spin up a remote development environment and complete end-to-end tasks with a single click, using Codex to edit and test changes. Codex automatically cleans up tech debt like dead code and expired experiments, improving code quality and reducing latency across codebases. It also takes on repetitive, well-understood changes, reducing backlog and significantly accelerating engineering velocity.

Availability and pricing updates

The Slack integration and Codex SDK are available to developers on ChatGPT Plus, Pro, Business, Edu, and Enterprise plans starting today, while the new admin features will be available to Business, Edu, and Enterprise. 

Starting October 20, Codex cloud tasks will also begin counting towards your Codex usage. Learn more about how Codex pricing works for each plan here⁠.

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

Source