OpenAI updates the o3-mini model, showing the "thinking chain" that gives the answer

近日，OpenAI 在推出其旗舰AI 模型o3和o3-mini 后，进行了重大更新，改变了o3-mini 的回应方式。现在，这款模型不仅能够回答用户的问题，还能展示其思考过程，为用户提供更多透明度。 This change marks an important step in improving user experience, making the use of AI tools more user-friendly.

The updated o3-mini, especially in the "high reasoning" mode, whether it is free users or paid users, can clearly see the "thinking chain" of the model before giving the answer. This thinking chain shows how the model understands the problem and reasons. For example, when the user asked "Why isn't today a Friday?", o3-mini not only gave a logical answer, but also analyzed the user's emotions, pointed out that this might be a humorous question, and even speculated that the user might be looking forward to the arrival of the weekend.

In addition, the new features of o3-mini include the ability to obtain data in real time from the network, which allows the model to attach source links when providing answers to improve the reliability of the answers. This advancement has made the integration of o3-mini in major enterprise applications, such as the combination of tools such as Microsoft's Azure OpenAI service and GitHub C opilot.

It is worth mentioning that compared with OpenAI's o3-mini, DeepSeek's R1 model shows a more complete reasoning process during thinking, and many AI researchers believe that this transparency is more ideal. While OpenAI does not fully demonstrate the reasoning steps of o3-mini, the company says it strives to find a balance point while protecting competitiveness so that users can understand the main thinking process of the model.

Overall, this update from OpenAI not only enhances the functions of o3-mini, but also allows users to feel a smoother and smart interactive experience during use. This change undoubtedly provides a new perspective and direction for the development of the AI ​​field.