OpenAI released GPT-5.4 mini and GPT-5.4 nano on March 17, 2026, adding two smaller and faster variants to its GPT-5.4 model family. The company positioned both models as purpose-built for the emerging agentic AI era, where applications need to coordinate multiple model calls at high speed and low cost.
GPT-5.4 mini brings much of the capability of the full GPT-5.4 model into a faster package. It scores 54.4% on SWE-Bench Pro, approaching the full model's 57.7%, and hits 72.1% on OSWorld-Verified for computer use tasks. OpenAI says mini runs more than 2x faster than GPT-5 mini while significantly improving on coding, reasoning, multimodal understanding, and tool use. It is available in the API, Codex, and ChatGPT, priced at $0.75 per million input tokens and $4.50 per million output tokens.
GPT-5.4 nano is the smallest and cheapest model in the lineup, designed for classification, data extraction, ranking, and lightweight subagent tasks. It is API-only at $0.20 per million input tokens and $1.25 per million output tokens. At these prices, a developer could describe 76,000 photos for roughly $52.
The release signals OpenAI's strategic focus on the agent orchestration layer. As AI applications increasingly rely on multi-model architectures where a primary model delegates subtasks to faster subordinate models, the economics of those subordinate calls become critical. Mini and nano are explicitly designed for that role.