<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
<channel>
  <title>MATRIX · DwarfStar ds4</title>
  <link>https://derekwang2002.github.io/ai-intelligence/en/projects/dwarfstar-ds4/</link>
  <description>Material changes in AI models, agents, engineering and infrastructure, with primary sources and recommendations.</description>
  <language>en</language>
  <lastBuildDate>Mon, 02 Nov 2026 16:56:00 GMT</lastBuildDate>
  <atom:link href="https://derekwang2002.github.io/ai-intelligence/en/feeds/projects/dwarfstar-ds4.xml" rel="self" type="application/rss+xml"/>
  <item>
    <title>DwarfStar builds a narrow local runtime for frontier open weights</title>
    <link>https://derekwang2002.github.io/ai-intelligence/en/events/ev-20261002-04/</link>
    <guid isPermaLink="false">ev-20261002-04-published</guid>
    <pubDate>Sat, 03 Oct 2026 16:01:57 GMT</pubDate>
    <category>New</category>
    <category>TRIAL</category>
    <description>&lt;p&gt;DwarfStar is an MIT-licensed C inference engine specialized for DeepSeek V4, GLM 5.x and Qwen3.8 on Metal, CUDA and ROCm. It combines project-specific GGUFs, SSD streaming, persistent KV caches, OpenAI- and Anthropic-style APIs and a native coding agent. The project has more than 23,000 GitHub stars but describes itself as fast-changing beta software.&lt;/p&gt;&lt;p&gt;Recommendation: TRIAL&lt;/p&gt;&lt;p&gt;Sources: &lt;a href=&quot;https://github.com/antirez/ds4&quot;&gt;DwarfStar ds4 repository&lt;/a&gt; · &lt;a href=&quot;https://dwarfstar.sh/&quot;&gt;DwarfStar ds4 documentation&lt;/a&gt;&lt;/p&gt;</description>
  </item>
</channel>
</rss>
