<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>Oxelan Signal — #open-weights</title>
    <link>https://blog.oxelan.com</link>
    <description>Posts tagged open-weights on AI, security and engineering.</description>
    <language>en-us</language>
    <lastBuildDate>Thu, 19 Nov 2026 03:38:12 GMT</lastBuildDate>
    <managingEditor>hello@oxelan.com (Kusal Tharindu)</managingEditor>
    <copyright>© 2026 Oxelan</copyright>
    <atom:link href="https://blog.oxelan.com/tags/open-weights/rss.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>Palo Alto Networks to buy Applied Compute for $4.5 billion</title>
      <link>https://blog.oxelan.com/security/palo-alto-networks-to-buy-applied-compute-for-4-5-billion/</link>
      <guid isPermaLink="true">https://blog.oxelan.com/security/palo-alto-networks-to-buy-applied-compute-for-4-5-billion/</guid>
      <pubDate>Thu, 19 Nov 2026 03:38:12 GMT</pubDate>
      <description>Palo Alto Networks agreed to acquire AI infrastructure company Applied Compute in an all-stock deal valued at $4.5 billion, and plans to offer its platform to security customers.</description>
      <category>open-weights</category>
      <category>inference</category>
    </item>
    <item>
      <title>Researchers took over the Kokoro model&apos;s Hugging Face account</title>
      <link>https://blog.oxelan.com/security/researchers-took-over-the-kokoro-model-s-hugging-face/</link>
      <guid isPermaLink="true">https://blog.oxelan.com/security/researchers-took-over-the-kokoro-model-s-hugging-face/</guid>
      <pubDate>Wed, 18 Nov 2026 02:39:53 GMT</pubDate>
      <description>According to BleepingComputer, researchers registered a domain listed in Kokoro&apos;s public DONATE.md file, reset the publisher&apos;s password, and then poisoned the model&apos;s weights in a local test.</description>
      <category>supply-chain</category>
      <category>open-weights</category>
      <category>authentication</category>
    </item>
    <item>
      <title>MiniMax releases H3-Image, an open-weight 4K image model</title>
      <link>https://blog.oxelan.com/ai/minimax-releases-h3-image-an-open-weight-4k-image-model/</link>
      <guid isPermaLink="true">https://blog.oxelan.com/ai/minimax-releases-h3-image-an-open-weight-4k-image-model/</guid>
      <pubDate>Tue, 10 Nov 2026 18:40:25 GMT</pubDate>
      <description>MiniMax released H3-Image, an open-weight image generation model that produces a full-resolution 4K image in under 10 seconds, with weights released and access through the MiniMax API Platform and Hailuo AI.</description>
      <category>open-weights</category>
      <category>inference</category>
    </item>
    <item>
      <title>Open-weight models: the bill arrives after the benchmark</title>
      <link>https://blog.oxelan.com/ai/open-weight-models-economics/</link>
      <guid isPermaLink="true">https://blog.oxelan.com/ai/open-weight-models-economics/</guid>
      <pubDate>Fri, 23 Oct 2026 00:00:00 GMT</pubDate>
      <description>Open weights remove a license fee, not the operating cost. The decision hinges on utilization, latency, privacy and the team you can staff.</description>
      <category>open-weights</category>
      <category>inference</category>
    </item>
  </channel>
</rss>
