OpenAI Brings GPT‑5.4 Mini and Nano to More Users

OpenAI has quietly expanded the GPT‑5.4 family with two smaller variants: GPT‑5.4 mini and GPT‑5.4 nano. Unlike the full GPT‑5.4 model that was positioned for heavy professional work, these lighter versions are aimed at broader access and cost‑sensitive use cases.

GPT‑5.4 mini is now accessible to Free and Go ChatGPT users by choosing the "Thinking" option in the model menu. For paid subscribers, mini also functions as the fallback when rate limits on the full GPT‑5.4 are reached. OpenAI says mini improves on its GPT‑5.0 mini predecessor in key areas such as reasoning, multimodal understanding and tool use. In practice, that means better parsing of images and audio and smarter handling of tasks like web search, all while running more than twice as fast than the older mini.

The other entrant, GPT‑5.4 nano, is designed primarily for developers and automated agents. OpenAI is offering nano exclusively through its API rather than in the chatbot interface. It's pitched as a fast, cost‑efficient option for jobs like data classification and extraction, where throughput matters more than the deepest contextual reasoning.

OpenAI has set nano's starting price at $0.20 per million input tokens, signaling an emphasis on affordability for high‑volume API workloads. The company expects developers to route simpler, repetitive tasks to nano, while reserving the larger GPT‑5.4 models for complex, high‑stakes work.

For users, the change broadens options: free users gain access to a noticeably smarter mini model, and developers get a compact API‑only model they can scale. It’s a pragmatic step by OpenAI to balance capability, cost and access as the ecosystem around large models continues to diversify.