OpenAI unveils GPT-5.3-Codex, which can tackle more advanced and complex coding tasks (www.techradar.com)

OpenAI has launched GPT-5.3-Codex, a significant upgrade that transforms its previous coding tool into a comprehensive computer-using agent. This new version is not only capable of writing and reviewing code but also executing complex tasks on a computer with enhanced reasoning abilities. Operating 25% faster than its predecessor, GPT-5.2, Codex 5.3 utilizes fewer tokens, allowing it to tackle intricate projects like game development and web applications more efficiently. Its ability to self-iterate with minimal human input marks a major leap in autonomous AI capabilities. Notably, GPT-5.3-Codex has set performance records across various benchmarks, including SWE-Bench and OSWorld, and is the first OpenAI model to assist in its own development—demonstrating its proficiency in debugging, deployment, and evaluation tasks. This evolution signifies a pivotal shift in AI, as Codex now integrates coding with practical computer operation, thereby enhancing productivity and collaboration in software development. The tool is available to all paid ChatGPT users across multiple platforms, further democratizing access to advanced AI capabilities for developers.