<?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>Claude on Loop &amp; Retry</title>
    <link>https://loopandretry.github.io/tags/claude/</link>
    <description>Recent content in Claude on Loop &amp; Retry</description>
    <image>
      <title>Loop &amp; Retry</title>
      <url>https://loopandretry.github.io/images/og-default.png</url>
      <link>https://loopandretry.github.io/images/og-default.png</link>
    </image>
    <generator>Hugo</generator>
    <language>en-us</language>
    <lastBuildDate>Thu, 20 Aug 2026 22:00:00 -0400</lastBuildDate>
    <atom:link href="https://loopandretry.github.io/tags/claude/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>What actually drives your Claude bill: cache misses, quadratic context, and prepaid retries</title>
      <link>https://loopandretry.github.io/posts/what-drives-your-claude-bill/</link>
      <pubDate>Thu, 20 Aug 2026 22:00:00 -0400</pubDate>
      <guid>https://loopandretry.github.io/posts/what-drives-your-claude-bill/</guid>
      <description>The per-token rate on Anthropic&amp;rsquo;s pricing page is real, but it&amp;rsquo;s the least useful number for predicting your actual invoice. Three multipliers move the bill far more than model choice does — a cache that silently misses, a transcript that grows quadratically with run length, and a best-of-N pattern that prepays for latency you may not need. Here&amp;rsquo;s the arithmetic on all three, combined, with real numbers.</description>
    </item>
  </channel>
</rss>
