OpenAI has expanded its model lineup with the release of GPT-5.4 Mini and GPT-5.4 Nano, two new variants that follow the earlier launch of GPT-5.4 this month, which was positioned primarily for professional use cases such as programming and data analysis. The new additions are designed to extend the reach of OpenAI’s latest generation of models to a broader range of users and use cases, from everyday consumers accessing ChatGPT for free to developers building cost-sensitive applications through the company’s application programming interface.
GPT-5.4 Mini is now available to Free and Go tier users within ChatGPT, accessible by selecting the Thinking option from the interface menu. For paid subscribers, the model functions as a fallback when usage limits for the full GPT-5.4 model are reached, ensuring continuity of service without a hard cutoff. OpenAI has stated that GPT-5.4 Mini delivers meaningfully improved performance over GPT-5.0 Mini across several dimensions, including reasoning capability, multimodal understanding, and tool usage. The model demonstrates enhanced ability to process non-text inputs such as images and audio, and brings improved performance on tasks involving web search. OpenAI also noted that GPT-5.4 Mini operates at more than twice the speed of its predecessor, a significant practical improvement for users who rely on the model for fast, iterative tasks.
GPT-5.4 Nano, on the other hand, is built for a different kind of deployment entirely. Rather than being available within the ChatGPT consumer interface, it is offered exclusively through OpenAI’s application programming interface and is optimised for tasks where speed and cost efficiency take priority over raw capability, such as data classification and information extraction at scale. The model is priced starting at 0.20 United States dollars per million input tokens, making it one of the more accessible options in OpenAI’s current portfolio for developers building agent-based workflows. OpenAI has indicated that larger models can be used to assign tasks to artificial intelligence agents powered by GPT-5.4 Nano, positioning the model as a lightweight but capable component within more complex multi-model architectures rather than a standalone general-purpose tool.
Follow the SPIN IDG WhatsApp Channel for updates across the Smart Pakistan Insights Network covering all of Pakistan’s technology ecosystem.
