GPT-5 nano (medium) is a lightweight configuration of OpenAI’s GPT-5 model family, released on August 7, 2025. It represents the smallest tier in the GPT-5 series, optimized for ultra-low latency and high-throughput applications. This specific configuration utilizes a medium reasoning effort setting, which balances the speed of the nano architecture with improved logical consistency and instruction following compared to the "minimal" effort mode.
The model features a unified multimodal architecture capable of processing text, images, and audio natively. It supports a context window of 400,000 tokens and a maximum output limit of 128,000 tokens. As part of OpenAI's adaptive model system, it incorporates hidden reasoning tokens that allow the model to think through instructions before generating final outputs, reducing hallucination rates compared to previous lightweight models.
GPT-5 nano (medium) is positioned as a successor to earlier compact models like GPT-4o-mini and GPT-4.1-nano. It is primarily used for tasks requiring rapid interactions, such as real-time chat assistants, basic classification, and high-volume summarization. While it has less reasoning depth than the larger GPT-5 or GPT-5 Pro variants, the medium reasoning setting provides a measurable accuracy boost for moderately complex workflows where cost-efficiency is the primary constraint.