Recently, OpenAI announced that it will delay the release of its first open-source large model, originally scheduled for next week. OpenAI CEO Sam Altman stated that the main reason for the delay is to conduct more comprehensive safety testing and review high-risk areas. He emphasized that once the model weights are made public, they cannot be recalled. This new model represents a new challenge for OpenAI, and they want to ensure everything is as perfect as possible.
Aidan Clark, OpenAI's Vice President of Research and the leader of this open-source project, also revealed that although the model performs "extraordinarily" in terms of capabilities, the company has very strict standards for open-source models. To ensure that the released model meets all expectations, the OpenAI team needs more time for detailed refinement. After all, once the model is released, it cannot be modified or withdrawn again, and such responsibility is no small matter.
According to previous reports, the performance of this new model is expected to be comparable to the existing o3-mini, and it may be named "Open Model." However, it should be noted that this name could cause confusion with the traditional concept of "open source." The extent to which the model is open will depend on whether the complete code, training details, and corresponding licenses are made public.
In today's technological environment, safety and responsibility are particularly important. Although OpenAI's decision may disappoint some eager users, it is a responsible move toward society and users. In the future, the release of this model is expected to bring more attention and anticipation regarding safety and practicality.
Although OpenAI's decision to delay the release is a disappointing message, in the long run, this decision will lay a more solid foundation for the model's security and pave the way for subsequent successful applications.