OpenAI introduced a new version of AI GPT 5.3 Codex, which is focused on automating software development. The model can work continuously for up to seven hours on projects of varying complexity and maintain context throughout the process. GPT 5.3 Codex demonstrates 30 percent higher accuracy in solving architectural tasks compared to previous versions, and the speed of query processing has increased by a quarter.
The AI successfully passes 95 percent of code security tests, increasing trust in automatic solutions in programming. Thanks to integration with work tools, GPT 5.3 Codex allows automating up to 40 percent of code review and repository audit processes. This helps teams reduce product release time and decrease the amount of routine work.
The model acts as a full-fledged agent that plans steps, works with tools, and operates in a real work environment. It is suitable not only for writing code but also for analytics, research, documentation creation, and tasks related to product development. OpenAI used Codex in its own development cycle, allowing the model to participate in creating itself.
The release of GPT 5.3 Codex indicates the growing capabilities of AI in the field of software development automation and the expansion of tasks that can be delegated to such systems. Users receive a tool that simplifies and accelerates work on complex projects and improves the quality of the final product.
