OpenAI's GPT-5.3-Codex model is rolling out to GitHub Copilot users across all subscription tiers. The new agentic coding model achieves higher benchmark scores and delivers up to 25% faster performance on agentic coding tasks compared to GPT-5.2-Codex, with improved reasoning for complex, tool-driven workflows. The model is available in VS Code, github.com, GitHub Mobile, CLI, and the Copilot Coding Agent, though Enterprise and Business administrators must enable it via policy settings.
Sort: