OpenAI has expanded the GPT-5.4 family with GPT-5.4 mini and GPT-5.4 nano, two smaller models designed for speed, efficiency, and high-volume AI workloads. Despite their more compact size, they are far from limited. Both are aimed at specific tasks where responsiveness, scalability, and lower costs matter most, especially in coding workflows and AI systems built around subagents. GPT-5.4 mini is already available in ChatGPT, while GPT-5.4 nano is currently limited to the API.
What GPT-5.4 mini and nano can do
According to OpenAI, GPT-5.4 mini is its strongest mini model so far for coding, computer use, and subagent-based workflows. It is optimized for tasks that require fast reasoning, multimodal understanding, and efficient tool use, making it well suited to environments where quick responses are critical. OpenAI says it delivers performance close to full GPT-5.4 on several benchmarks, while running faster and at lower cost.
That makes GPT-5.4 mini particularly useful in systems where models of different sizes work together. In workflows like Codex, for example, a larger model can handle planning, coordination, and final decision-making, while GPT-5.4 mini takes care of narrower tasks in parallel, such as searching through a codebase, reviewing large files, or processing supporting documents. OpenAI also highlights its strength in multimodal tasks, including the interpretation of screenshots and dense user interfaces.
GPT-5.4 nano is the smallest and cheapest model in the GPT-5.4 lineup. It is designed for lightweight, high-volume use cases where cost efficiency is the priority, including classification, data extraction, ranking, and simple subagent support tasks. While it is less powerful than mini, it is meant to offer a practical option for developers who need speed and scale without the higher cost of larger models.
As for availability, GPT-5.4 mini is already offered through the API, Codex, and ChatGPT. In ChatGPT, OpenAI says it is available to Free and Go users through the Thinking option in the plus menu, while other users may encounter it as a fallback when they hit limits on GPT-5.4 Thinking. GPT-5.4 nano, on the other hand, is available only through the API. OpenAI lists pricing at $0.75 per million input tokens and $4.50 per million output tokens for GPT-5.4 mini, and $0.20 input / $1.25 output per million tokens for GPT-5.4 nano.
