
  <rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
    <channel>
      <title>Isomoes Blog</title>
      <link>https://isomoes.github.io/en/blog</link>
      <description>Isomoes personal blog</description>
      <language>en-us</language>
      <managingEditor>jiahaoxiang2000@gmail.com (Isomoes)</managingEditor>
      <webMaster>jiahaoxiang2000@gmail.com (Isomoes)</webMaster>
      <lastBuildDate>Sat, 12 Sep 2026 00:00:00 GMT</lastBuildDate>
      <atom:link href="https://isomoes.github.io/en/tags/harness/feed.xml" rel="self" type="application/rss+xml"/>
      
  <item>
    <guid>https://isomoes.github.io/en/blog/tools/model-harness-fit</guid>
    <title>Same Model, Different Harness: Why Does Performance Change?</title>
    <link>https://isomoes.github.io/en/blog/tools/model-harness-fit</link>
    <description>Why can the same GPT model stop midway, slow down, or struggle to finish a task in a different environment? Drawing on our experience with OpenCode and Codex, we examine how the harness affects availability and task performance, and how to choose a better fit.</description>
    <pubDate>Sat, 12 Sep 2026 00:00:00 GMT</pubDate>
    <author>jiahaoxiang2000@gmail.com (Isomoes)</author>
    <category>ai-agent</category><category>harness</category><category>codex</category><category>opencode</category><category>deepseek-harness</category><category>tool-use</category>
  </item>

    </channel>
  </rss>
