<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Engineering on homeio.org</title><link>https://homeio.org/topics/engineering/</link><description>Recent content in Engineering on homeio.org</description><generator>Hugo</generator><language>en-us</language><lastBuildDate>Fri, 13 Nov 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://homeio.org/topics/engineering/feed.xml" rel="self" type="application/rss+xml"/><item><title>Mellum 2.1 12B, Flint 40B and Arete 27B: five small coding fixes</title><link>https://homeio.org/posts/2026/2026-11-13-local-coding-agents-real-project-benchmark/</link><pubDate>Fri, 13 Nov 2026 00:00:00 +0000</pubDate><guid>https://homeio.org/posts/2026/2026-11-13-local-coding-agents-real-project-benchmark/</guid><description>&lt;p&gt;I wanted to know whether a local coding agent could handle small, well-defined
changes without me stepping in. Not build an application from scratch: read an
existing module, fix one behavior, check it, and stop.&lt;/p&gt;
&lt;p&gt;I gave Mellum 2.1 12B-A2.5B, Flint 40B-A2.5B and Arete 27B the same five tasks
from my Julia, Python and Ruby projects. All three used &lt;strong&gt;OMP 18.20.0&lt;/strong&gt; as the
coding-agent harness, with the same limits. &lt;strong&gt;Arete was the only model to finish all five successfully.&lt;/strong&gt;
Mellum was usually faster when it succeeded. Flint sometimes had the right code
on disk but kept working until it ran out of requests.&lt;/p&gt;</description></item></channel></rss>