GitHub Copilot Student Plan Removes GPT-5.3-Codex from Manual Model Picker in Favor of Auto Selection

GitHub has announced an update to the Copilot Student plan that modifies how users interact with specific large language models. Specifically, the GPT-5.3-Codex model is being removed from the manual model selection interface. This change aims to streamline the user experience by relying on internal logic for model routing rather than manual configuration by the student developer. While the model is no longer visible in the picker, it remains accessible through the auto model selection feature. This system is designed to intelligently match the most appropriate model to the specific task or context at hand. Developers who previously relied on manual selection for specific performance characteristics should now monitor how the automated system handles their typical code generation requests. Operational teams and students should focus on evaluating input and output compatibility during this transition. Since the specific model version might vary based on the auto-selection logic, it is important to re-verify any automated testing or evaluation frameworks. Testing for regressions in edge cases or niche programming languages where Codex performed uniquely is recommended to ensure consistent code quality. To minimize disruption, it is advisable to audit existing development environments and identify any hard dependencies on manual model toggling. Gradually transitioning to the automated selection logic allows teams to assess performance impacts without sudden failures. Ensuring that local configurations do not attempt to force a deprecated manual selection will prevent UI inconsistencies or unexpected behavior in the Integrated Development Environment.