OpenAI recently unveiled GPT-5.4 mini and nano, extending the capabilities of its latest AI models. Following the March launch of GPT-5.4, primarily aimed at professionals for tasks like programming and data analysis, the company is now making a version of its advanced technology accessible to a broader audience.
Significantly, GPT-5.4 mini is now available to ChatGPT Free and Go users, bringing a portion of GPT-5.4’s intelligence to them. Users can access this model by selecting ‘Thinking’ from ChatGPT’s plus menu. For paid subscribers, GPT-5.4 mini acts as a dependable fallback when their usage limits for the full GPT-5.4 model are reached.
OpenAI highlights that GPT-5.4 mini surpasses its predecessor, GPT-5.0 mini, across several critical domains, including reasoning, multimodal understanding, and tool utilization. This means the model is more adept at interpreting diverse inputs like images and audio, and possesses a more sophisticated grasp of functions such as web searching. Furthermore, GPT-5.4 mini operates at more than double the speed of its earlier version.
In contrast, GPT-5.4 nano is engineered for specific applications, such as data classification and extraction, where efficiency and cost-effectiveness are paramount. This model will not be integrated into the ChatGPT chatbot but will instead be exclusively available through OpenAI’s API service. The vision is for developers to leverage more powerful models to delegate tasks to AI agents powered by GPT-5.4 nano. Reflecting its specialized role, GPT-5.4 nano is priced starting at $0.20 per million input tokens.
