Announcing the launchGPT-4.1 ModelLater, OpenAI launchedNew o3 and o4-mini modelsThe former is touted as its highest-performing artificial intelligence model to date, while the latter targets lower-cost applications but emphasizes that it still performs quite well in the same application areas.
In addition to touting the inference capabilities of the two models, OpenAI also explained that both o3 and 04-mini can use all of ChatGPT's features in a multimodal manner, including web browsing and image generation, to more efficiently solve multi-step problems and thus be able to handle various tasks more independently.
Both models also support image visual analysis capabilities, and can analyze and interpret the images they "see" while making inferences. OpenAI expects that artificial intelligence will be able to significantly improve performance and practicality when it has better inference capabilities and can access more functions.
On the other hand, OpenAI also announced the launch of a new coding agent function called Codex CLI, which is expected to allow developers to connect OpenAI's artificial intelligence models with localized program code through this function, thereby creating more corresponding artificial intelligence agent services. Currently supported models include o3 and o4-mini officially announced this time, and support for GPT-4.1 is also being prepared.
As for the previous market speculationGPT-5 artificial intelligence model, OpenAI CEO Sam Altman revealed that it will be announced in the next few months and is expected to be launched this year.
The o3 and o4-mini models announced this time will be available first to users who subscribe to ChatGPT Plus, ChatGPT Pro or ChatGPT Team. Users can upgrade to the more powerful version o3-pro in the next few weeks.
