OpenAI has unveiled two new models, o3 and o4-mini, which integrate advanced simulated reasoning capabilities with access to various tools such as web browsing and coding. These models represent a significant step forward in ChatGPT’s abilities, offering better cost efficiency and improved performance. While o3 focuses on complex analysis, o4-mini prioritizes speed and cost-effectiveness. Both models enable users to solve multi-step problems by autonomously utilizing aids like searching for data, writing Python code, and generating visual graphs. They also feature multimodal capabilities, allowing them to interpret and “think with” images. Early feedback indicates strong performance, though independent verification of benchmarks is still needed. Pricing for the models reflects a reduction compared to predecessors, making them more accessible for developers. Additionally, OpenAI introduced Codex CLI, an experimental terminal application designed to enhance coding tasks. — new from Ars Technica
