<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:media="http://search.yahoo.com/mrss/">
  <channel>
    <title>Bramble: Muse Wonder</title>
    <link>https://news.ketelle.co.uk/entities/muse-wonder</link>
    <description>Stories involving Muse Wonder, told neutrally, with how each side covers them.</description>
    <language>en-gb</language>
    <atom:link href="https://news.ketelle.co.uk/entities/muse-wonder/feed.xml" rel="self" type="application/rss+xml"/>
    <item>
      <title>Meta launches Muse Wonder and Muse Wonder Full AI models, claims parity with OpenAI and Anthropic</title>
      <link>https://news.ketelle.co.uk/stories/meta-launches-muse-wonder-and-muse-wonder-full-ai-models</link>
      <guid isPermaLink="true">https://news.ketelle.co.uk/stories/meta-launches-muse-wonder-and-muse-wonder-full-ai-models</guid>
      <pubDate>Mon, 12 Oct 2026 21:11:12 +0100</pubDate>
      <category>Technology</category>
      <description>Meta's Superintelligence Labs released Muse Wonder and Muse Wonder Full on 12 October, which Meta says rival the best models from OpenAI and Anthropic. The benchmark claims come from Meta's own testing. (6 outlets: 2 left · 2 centre · 2 right)</description>
      <media:content url="https://platform.theverge.com/wp-content/uploads/sites/2/2026/10/STKB394_MUSE_AI_CVIRGINIA_D.png?quality=90&amp;strip=all&amp;crop=0%2C9.9676601489831%2C100%2C80.064679702034&amp;w=1200" medium="image"/>
    </item>
  </channel>
</rss>
