Codex bundles LibreOffice

OpenAI's Codex desktop application, now rebranded as ChatGPT, has been discovered to bundle significant third-party software components within its installation package. A user investigating cached files found that the application contains approximately 1.7GB of bundled resources, including complete Python and Node.js installations alongside LibreOffice components. This discovery raises important questions about application bloat, transparency, and resource management in AI-powered development tools.

The Codex desktop application maintains a dedicated folder in the system cache directory called "codex-primary-runtime" that contains substantially more than the core AI application itself. Among the bundled components are full programming language installations and office productivity software. This approach—while potentially simplifying dependency management and ensuring compatibility across different user systems—significantly increases the application's disk footprint and resource requirements. Users downloading the ChatGPT desktop application are essentially receiving multiple standalone applications without explicit notification of what additional software is being installed on their systems.

The bundling strategy suggests OpenAI's focus on creating a self-contained, cross-platform experience where users don't encounter missing dependencies or version conflicts. However, this comes at the cost of redundancy for users who already have these tools installed independently.

• Transparency concerns: Users may not realize what additional software is being installed with downloaded applications
• Storage efficiency: Bundling creates significant disk waste for users with existing Python, Node.js, or LibreOffice installations
• System resource management: Multiple copies of runtime environments consume additional memory and processing power
• Installation practices: Raises questions about best practices for modern AI application distribution
• User control: Limited ability for users to opt out of bundled components they don't need

This discovery highlights a growing tension in software distribution between convenience and resource efficiency. As AI tools become more integrated into developer workflows, understanding what's actually being installed on systems becomes increasingly important. The trend toward bundled, monolithic applications may prioritize user experience but can significantly impact system performance and storage capacity, particularly for users managing multiple development environments or working on resource-constrained machines.