<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/"><channel><title>成本曲线 on AI 实战派 · 从技术到赚钱</title><link>https://guijiagi.com/tags/%E6%88%90%E6%9C%AC%E6%9B%B2%E7%BA%BF/</link><description>Recent content in 成本曲线 on AI 实战派 · 从技术到赚钱</description><generator>Hugo</generator><language>zh-cn</language><copyright>本站内容采用 CC BY-NC-SA 4.0 国际许可协议授权</copyright><lastBuildDate>Fri, 09 Oct 2026 15:18:00 +0800</lastBuildDate><atom:link href="https://guijiagi.com/tags/%E6%88%90%E6%9C%AC%E6%9B%B2%E7%BA%BF/index.xml" rel="self" type="application/rss+xml"/><item><title>推理时计算的边际收益曲线：2026 年还要不要堆「思考时长」</title><link>https://guijiagi.com/posts/2026-10-09-04-test-time-compute-marginal/</link><pubDate>Fri, 09 Oct 2026 15:18:00 +0800</pubDate><guid>https://guijiagi.com/posts/2026-10-09-04-test-time-compute-marginal/</guid><description>o 系列开启的 test-time compute scaling 在 2026 年出现边际递减迹象。本文拆解「多想一步」的成本与正确率换算。</description></item></channel></rss>