OpenAI has started testing a new 'reasoning' AI model family, o3 and o3-mini, as it competes with tech giants like Google.o3 is the successor to o1 and is designed for complex programming, maths, and science tasks.The models spend time analyzing their responses before answering queries, making them reliable but computationally expensive.OpenAI plans to launch o3-mini by the end of January and full o3 afterwards, with safety researchers able to preview o3-mini.