
  <rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
    <channel>
      <title>tinapark.dev</title>
      <link>https://tinapark.dev/blog</link>
      <description>AI engineer and community builder. Building AI apps, hosting VisibleBuilders cowork meetups, and writing about what I learn shipping in public.</description>
      <language>en-us</language>
      <managingEditor>tinapark.business@gmail.com (Tina Park)</managingEditor>
      <webMaster>tinapark.business@gmail.com (Tina Park)</webMaster>
      <lastBuildDate>Mon, 22 Jun 2026 00:00:00 GMT</lastBuildDate>
      <atom:link href="https://tinapark.dev/tags/gemma/feed.xml" rel="self" type="application/rss+xml"/>
      
  <item>
    <guid>https://tinapark.dev/blog/2026-06-20-gemma4</guid>
    <title>From 147 Seconds to 3: How Gemma 4 Gets Fast Enough to Run on a Laptop</title>
    <link>https://tinapark.dev/blog/2026-06-20-gemma4</link>
    <description>At Google I/O Extended Newport Beach, I watched Gemma 4 go from a 147-second response down to a few seconds — with no new hardware, just five inference optimization techniques stacked on top of each other. Here is what each one actually does, in plain language.</description>
    <pubDate>Mon, 22 Jun 2026 00:00:00 GMT</pubDate>
    <author>tinapark.business@gmail.com (Tina Park)</author>
    <category>AI</category><category>LLM</category><category>Gemma</category><category>GoogleIO</category><category>OnDeviceAI</category><category>MLX</category><category>Optimization</category>
  </item>

    </channel>
  </rss>
