OpenAI launches GPT-5-Codex with a 74.5% success rate on real world coding

OpenAI's new GPT-5-Codex achieves a 74.5% success rate on real-world coding tasks, significantly outperforming previous models. The purpose-built coding assistant uses 93.7% fewer tokens for lightweight tasks while spending more time on complex problems, provides transparency through logs and test results, and integrates with terminals, IDEs, GitHub and iOS. Available to Plus, Pro, Business, Edu and Enterprise users, this tool excels at code reviews, bug detection, and even frontend work with visual capabilities.