OpenAI recently released its latest small models — GPT-5.4mini and GPT-5.4nano. These two models are specifically designed for high-frequency tasks that require fast responses, marking another leap in AI technology.

According to OpenAI's official announcement, GPT-5.4mini and nano are optimized based on the previous GPT-5.4 model, retaining many of its advantages while improving speed and efficiency. Especially in application scenarios such as code writing, logical reasoning, and multimodal understanding, these small models have shown better performance than traditional large models. The GPT-5.4mini runs twice as fast as the previous version, allowing users to complete tasks more quickly in high-pressure work environments.

image.png

In particular, GPT-5.4mini excels in the rapid iteration of code workflows, efficiently handling complex tasks such as precise editing, codebase navigation, and front-end generation. In the multimodal field, GPT-5.4mini can quickly analyze dense user interface screenshots and complete various computer operations, further enhancing the user experience.

GPT-5.4nano is currently the smallest and most cost-effective version, designed for users who prioritize economy and speed. As a major upgrade of GPT-5nano, it performs well in text classification, data extraction, and simple assistance tasks, making it an ideal choice for developers.

image.png

These two new models are now fully available. GPT-5.4mini can be accessed through API, Codex, and ChatGPT, while GPT-5.4nano is mainly provided through API. More excitingly, their costs are very affordable, greatly lowering the entry barrier for developers.

OpenAI's recent release undoubtedly brings new vitality to the AI field. Future intelligent application scenarios will be more diverse, and users' work efficiency will also be significantly improved. We look forward to the performance of these two models in practical applications, which will bring us more surprises!