OpenAI founder Sam Altman has unveiled the latest iteration of the Codex model, GPT-5.3-Codex, which promises significant improvements in coding performance. According to Jin10, the new model boasts enhanced controllability and real-time update capabilities during task execution. It requires only half the tokens compared to its predecessor, GPT-5.2-Codex, while increasing single token processing speed by over 25%. Additionally, GPT-5.3-Codex is designed to efficiently utilize computer resources, marking a substantial advancement in AI-driven coding solutions.