<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
<channel>
  <title>MATRIX · LightOnOCR-3</title>
  <link>https://derekwang2002.github.io/ai-intelligence/en/projects/lightonocr-3/</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/lightonocr-3.xml" rel="self" type="application/rss+xml"/>
  <item>
    <title>LightOn releases OCR models that extract layout, charts and images</title>
    <link>https://derekwang2002.github.io/ai-intelligence/en/events/ev-20261008-04/</link>
    <guid isPermaLink="false">ev-20261008-04-published</guid>
    <pubDate>Sun, 11 Oct 2026 16:02:40 GMT</pubDate>
    <category>New</category>
    <category>TRIAL</category>
    <description>&lt;p&gt;LightOn released 0.8B, 1B and 4B LightOnOCR-3 weights. The models convert pages to Markdown; a grounding mode also emits labeled bounding boxes, image descriptions and chart data. The Apache 2.0 weights permit commercial use. Reproduction scripts are public, but the reported scores remain vendor-run.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Why read: &lt;/strong&gt;Open-weight OCR, grounding and chart parsing provide an engineering candidate that can be tested directly on local documents.&lt;/p&gt;&lt;p&gt;Recommendation: TRIAL&lt;/p&gt;&lt;p&gt;Sources: &lt;a href=&quot;https://huggingface.co/blog/LightOnOCR/lightonocr-3&quot;&gt;LightOnOCR-3&lt;/a&gt; · &lt;a href=&quot;https://huggingface.co/LightOnAI/LightOnOCR-3-4B-1025&quot;&gt;LightOnOCR-3-4B-1025&lt;/a&gt;&lt;/p&gt;</description>
  </item>
</channel>
</rss>
