OpenAI merges Codex and ChatGPT: the live code assistant arrives

What happened?

On July 9, 2026, OpenAI held a live stream titled “The next chapter for ChatGPT”, where it presented the integration of Codex, its artificial intelligence engine for code generation and execution, directly into the main ChatGPT application. According to 9to5Mac, the company had already hinted at this move in June 2026, when it mentioned plans to incorporate Codex workflows into ChatGPT across all platforms.

The new feature allows users to request programming tasks in natural language, and ChatGPT not only generates the code but also executes it in a secure sandbox environment, displays results, identifies errors, and suggests fixes. Additionally, it integrates with GitHub repositories and deployment platforms like Vercel and Netlify. This integration represents the culmination of a process OpenAI began in 2021 with the launch of Codex as a closed beta, based on GPT-3. At that time, Codex proved capable of translating English descriptions into code in over a dozen languages, but its use required a separate API and technical knowledge. Now, by integrating it into ChatGPT, OpenAI removes that barrier, allowing any user with a subscription to access AI-assisted programming capabilities from a conversational chat.

Why is it important?

Codex was originally a standalone product, launched in 2021 as a beta, that allowed developers to generate code from English descriptions. However, its mass adoption was limited by the need to use a separate API. By integrating it into ChatGPT, OpenAI democratizes access to AI-assisted programming. Any user, even without deep technical knowledge, can now create scripts, analyze data, or prototype applications. According to data from the stream, a demo showed how a user with no prior experience could create a simple web application in minutes, simply by describing what they wanted. This contrasts with the approach of GitHub Copilot, which since its launch in 2021 as an extension for code editors, has been aimed primarily at developers already familiar with integrated development environments (IDEs). Copilot, based on GPT-4, offers real-time suggestions but does not execute code or provide a sandbox environment. The integration of Codex into ChatGPT goes a step further by offering a complete cycle: description, generation, execution, and debugging.

This move also responds to the growing competition from code assistants like GitHub Copilot (based on GPT-4) and no-code/low-code tools. By unifying the conversational experience with code execution, OpenAI is betting on a comprehensive assistant that directly competes with platforms like Replit and Google Colab. Replit, for example, already offered a browser-based development environment with AI capabilities, but its approach remained more technical. Google Colab, on the other hand, is oriented toward Python notebooks and does not offer a conversational interface. OpenAI's strategy recalls when they launched ChatGPT in 2022, integrating the language model into an accessible chat, which skyrocketed its adoption. Now, with Codex integrated, they seek to repeat that success in the programming domain.

What consequences will it have?

For developers

The barrier to entry for programming is drastically reduced. Junior developers can use ChatGPT to learn and debug, while senior developers can delegate repetitive tasks. However, there is a risk of over-reliance and potential loss of fundamental skills. A 2025 Stanford University study suggested that developers using AI assistants tend to make more security errors if they do not review the generated code. Additionally, the integration with GitHub allows ChatGPT to access existing repositories, facilitating code review and pull request creation, but also raises questions about the intellectual property of the generated code, especially if it is based on code with restrictive licenses.

For businesses

Companies will be able to accelerate development cycles, reduce training costs, and allow multidisciplinary teams to prototype solutions without waiting for engineers. However, concerns arise about security, intellectual property of generated code, and the need for human oversight. According to a 2026 Gartner report, companies that adopt AI coding assistants could reduce development time by 40%, but will also face security risks if they do not implement rigorous reviews. Integration with Vercel and Netlify enables direct deployments, speeding up production, but can also introduce vulnerabilities if the code is not audited.

For the labor market

Automation of coding tasks could displace basic programming jobs, but will also create demand for roles that integrate AI with ethical oversight and systems architecture. A 2025 McKinsey study estimated that up to 30% of routine coding tasks could be automated by 2030, but demand for developers with skills in AI, security, and systems design will increase by 20%. Furthermore, the integration of Codex into ChatGPT could accelerate the trend toward “citizen development,” where non-technical employees create software solutions, potentially changing the dynamics of IT departments.

What should readers know?

• Availability: The feature is rolling out gradually to ChatGPT Plus and Enterprise subscribers, with plans to reach free users in subsequent months. According to the stream, Plus users will have priority access, while Enterprise can customize the sandbox environment and integrations.
• Limitations: The execution environment is limited; it does not replace a full IDE for large projects. OpenAI warns that generated code should always be reviewed. In the demo, the sandbox had resource restrictions (CPU, memory, runtime) to prevent abuse. Applications requiring access to external databases or system files cannot be executed.
• Privacy: Executed code is processed on OpenAI servers; companies must evaluate compliance implications, especially with regulations like GDPR or CCPA. OpenAI has stated that it does not use Enterprise customers' code to train models, but free and Plus users must accept terms that allow data use for service improvement.
• Comparison with competitors: Unlike GitHub Copilot, which integrates into IDEs and offers real-time suggestions, ChatGPT with Codex provides a complete execution and debugging environment. However, Copilot has an advantage in integration with existing development workflows. Replit, on the other hand, offers a more complete development environment but lacks ChatGPT's conversational interface.

“We are witnessing the birth of a new paradigm: the conversational assistant that not only talks but does. The integration of Codex into ChatGPT marks a before and after in the accessibility of programming.” — Analyst at TheVortiq.

In summary, the integration of Codex into ChatGPT not only expands the assistant's capabilities but redefines how one can interact with programming. By lowering the technical barrier, OpenAI opens the door to a new generation of software creators, but also poses challenges that the industry must address in terms of security, ethics, and training. Time will tell whether this bet consolidates ChatGPT as the dominant code assistant or if competition manages to offer more specialized alternatives.