<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
<channel>
  <title>MATRIX · Strata</title>
  <link>https://derekwang2002.github.io/ai-intelligence/en/projects/strata/</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/strata.xml" rel="self" type="application/rss+xml"/>
  <item>
    <title>Strata runs Qwen3.8-Flash-Next on 12 GB gaming GPUs</title>
    <link>https://derekwang2002.github.io/ai-intelligence/en/events/ev-20260924-12/</link>
    <guid isPermaLink="false">ev-20260924-12-published</guid>
    <pubDate>Sun, 04 Oct 2026 16:02:19 GMT</pubDate>
    <category>New</category>
    <category>TRIAL</category>
    <description>&lt;p&gt;Strata is an MIT-licensed inference engine that distributes Qwen3.8-Flash-Next experts across GPU, RAM, CPU and SSD. It exposes OpenAI- and Anthropic-compatible local endpoints and documents a 12 GB GPU, 32 GB RAM and about 80 GB SSD floor. Author benchmarks report 60–94 output tokens per second on recent gaming GPUs.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Why read: &lt;/strong&gt;Strata brings a very large sparse model to consumer hardware with strong adoption.&lt;/p&gt;&lt;p&gt;Recommendation: TRIAL&lt;/p&gt;&lt;p&gt;Sources: &lt;a href=&quot;https://github.com/Niko1221/Strata&quot;&gt;Strata local MoE inference engine&lt;/a&gt;&lt;/p&gt;</description>
  </item>
</channel>
</rss>
