gpt-4.1-nano

GPT-4.1-nano belongs to the same family line as GPT-4.1 and GPT-4.1-mini, introduced as the smallest and most economical member of that series. It inherits the GPT-4.1 family's core strengths, including multimodal input handling for both text and images, long-context understanding, improved instruction following, and enhanced coding capabilities. Its design goal is to deliver that capability set at substantially reduced latency and cost, making it well suited for real-time applications and large-scale deployment of parallel or chained model calls, such as powering chatbots, coding copilots, and AI agents that need to invoke multiple lightweight models quickly.

Benchmark evidence positions GPT-4.1-nano as surprisingly capable for its size, scoring 80.1% on MMLU, 50.3% on GPQA, and 9.8% on Aider polyglot coding, which exceeds the performance of GPT-4o mini. The model operates within a one-million-token context window and supports structured output generation alongside parallel function calling for tool-based workflows. These characteristics make it a strong practical fit for lightweight, high-frequency workloads like classification, autocompletion, and other behind-the-scenes tasks where speed and cost matter more than top-tier reasoning depth.