Moments after Sam Altman took to social media to express his gratitude to developers for crafting complex code “character-by-character”, OpenAI introduced two new lightweight AI models crafted for the coding community, GPT-5.4 Mini and GPT-5.4 Nano. These models are aimed at delivering faster performance and lower costs for developers and businesses. These models now complete OpenAI’s most advanced GPT-5.4 family, which also includes the Pro, Thinking and the flagship model.
The new models are designed for speed-sensitive and cost-conscious environments. GPT-5.4 Mini is said to be well-suited for coding assistants, debugging tools, chatbots, and real-time AI systems that require both accuracy and responsiveness.
GPT-5.4 Nano targets backend and large-scale automation tasks such as data processing, content classification, and workflow orchestration. Its low cost makes it ideal for high-volume operations.
OpenAI states that these models promise to allow developers to build layered AI systems that optimise both performance and cost, marking a significant step toward more scalable and efficient AI deployment.
GPT-5.4 Mini and Nano: What do these new models offer?
GPT-5.4 Mini is positioned as a powerful yet efficient version of the flagship GPT-5.4 model. It aims to significantly improve coding, reasoning, and multimodal capabilities while running more than twice as fast as its predecessor, GPT-5 Mini. As a result, benchmarks show notable gains in software engineering and reasoning tasks, bringing it closer to flagship-level performance.
On the other hand, GPT-5.4 Nano is the smallest and most cost-efficient model in the lineup. It is designed for simpler, high-frequency tasks such as classification, ranking, and data extraction. While less powerful than Mini, Nano offers a strong balance between speed and performance for background operations.
Both models are optimised for “sub-agent workflows,” where multiple AI systems collaborate; larger models handle complex reasoning while smaller ones execute specific tasks.
How to access the GPT-5.4 Mini and Nano models
OpenAI has priced the new models to appeal to developers running large-scale applications. GPT-5.4 Mini costs approximately $0.75 per million input tokens and $4.50 per million output tokens.
GPT-5.4 Nano is even cheaper, priced at $0.20 per million input tokens and $1.25 per million output tokens, making it the most affordable option in the GPT-5.4 family.
As far as availability is concerned, GPT-5.4 Mini is accessible in ChatGPT (including Free and Go tiers via the “Thinking” feature), as well as through the API. Nano, however, is primarily available via API for developers building scalable applications.