<?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>Tool Calling on 梦兽编程</title><link>https://rexai.top/en/tags/tool-calling/</link><description>Recent content in Tool Calling on 梦兽编程</description><generator>Hugo -- 0.163.3</generator><language>en</language><copyright>梦兽编程</copyright><lastBuildDate>Mon, 27 Jul 2026 22:30:00 +0800</lastBuildDate><atom:link href="https://rexai.top/en/tags/tool-calling/index.xml" rel="self" type="application/rss+xml"/><item><title>Newer Claude Models Are Worse at Using Your Tools: An Engineering Trap of RL Training Backfire</title><link>https://rexai.top/en/posts/2026-07-28-better-models-worse-tools/</link><pubDate>Mon, 27 Jul 2026 22:30:00 +0800</pubDate><guid>https://rexai.top/en/posts/2026-07-28-better-models-worse-tools/</guid><description>Armin discovered that Opus 4.8 and Sonnet 5 are actually more error-prone than older models when calling Pi&amp;#39;s custom editing tool — because RL training overfits them to Claude Code&amp;#39;s built-in tools. This is a wake-up call for every third-party coding harness.</description></item></channel></rss>