OpenAI has released a new model called GPT-5.3-Codex, which is described as an advanced codex model for agent-style development and an agentic coding upgrade. The release of GPT-5.3-Codex is facing scrutiny over potential violations of California's AI safety law, with a watchdog alleging high-risk violations. The model's capabilities and potential impact are being evaluated, with some outlets comparing it to other models, such as Claude Opus 4.6, and noting varying assessments of its performance and safety.