OpenAI Launches Lightweight GPT-5.4 Mini and Nano Models for Low-Cost AI Agents

OpenAI has continued to broaden its model lineup in recent years, complementing its large flagship models with smaller, low-latency and lower-cost models for enterprise automation. GPT-5.4 Mini and Nano target coding, tool calling and multi-agent collaboration, allowing a primary agent to delegate large volumes of subtasks to less expensive models and lowering the computing barrier to deploying AI agents at scale.

OpenAI launched GPT-5.4 Mini and Nano in July 2026, saying they run about twice as fast as their predecessors and are optimized for coding and sub-agent workloads. The company also cut API prices for both models, with Nano starting at just $0.20 per million tokens. The models target high-frequency, batch-processing and multi-agent automation applications.