OpenAI unveils reasoning AI

OpenAI has launched new AI models designed to mimic human reasoning for complex coding and visual tasks. The models, o3 and o4-mini, represent a significant upgrade in ChatGPT's capabilities, enabling it to handle multifaceted questions more effectively. These models can utilise all ChatGPT tools, including web searching, Python analysis, and image generation, to provide detailed and thoughtful answers. OpenAI's o3 model sets a new standard in coding, maths, science, and visual perception, outperforming previous models on benchmarks like Codeforces and SWE-bench. It excels in analysing images, charts and graphics, making 20% fewer errors than o1 on real-world tasks, particularly in programming and business. The o4-mini offers a smaller, faster alternative. These releases aim to maintain OpenAI's competitive edge amidst growing competition.