2026年3月18日AgentsAPICoding

GPT-5.4 Mini and Nano: OpenAI's Models Built for the Subagent Era

OpenAI released GPT-5.4 mini and GPT-5.4 nano on March 17, 2026 — two smaller, faster models purpose-built for high-volume agentic workloads and subagent orchestration.

GPT-5.4 mini brings most of GPT-5.4's strengths to a model that runs over 2x faster. It approaches the full GPT-5.4 on SWE-Bench Pro and OSWorld-Verified, making it suitable as the primary model in coding agent pipelines. GPT-5.4 nano is the smallest and cheapest variant, priced at $0.20 per million input tokens and $1.25 per million output tokens, optimized for classification, data extraction, ranking, and supporting subagent tasks.

This release signals a significant shift in how OpenAI thinks about model deployment. Rather than one model to rule them all, the mini/nano tier explicitly targets the emerging pattern of multi-agent systems where a primary agent delegates to cheaper, faster subagents for routine work. GPT-5.4 nano's pricing makes it economically viable to run hundreds of parallel subagents.

GPT-5.4 mini is available in ChatGPT today. GPT-5.4 nano is API-only.

https://openai.com/index/introducing-gpt-5-4-mini-and-nano/
← 上一篇
discli:让AI代理从终端完全访问 Discord 的命令行工具
下一篇 →
GPT-5.4 Mini 与 Nano:OpenAI 为子智能体时代打造的模型
← 返回所有文章

评论

加载中...
>_