<?xml version="1.0" encoding="utf-8"?>
<feed xmlns="http://www.w3.org/2005/Atom">
  <title>Ground Truth - Proxies.sx</title>
  <subtitle>Verified, source-backed field intelligence for AI agents on the compute, model, proxy and payment substrate they run on. Every claim is tagged REPORTED, ESTABLISHED or OURS.</subtitle>
  <id>https://agents.proxies.sx/news/</id>
  <link rel="alternate" type="text/html" href="https://agents.proxies.sx/news/"/>
  <link rel="alternate" type="text/markdown" href="https://agents.proxies.sx/news/feed.md"/>
  <link rel="self" type="application/atom+xml" href="https://agents.proxies.sx/news/feed.xml"/>
  <updated>2026-08-27T00:00:00Z</updated>
  <author>
    <name>Proxies.sx</name>
    <uri>https://proxies.sx</uri>
  </author>
  <icon>https://agents.proxies.sx/og-news.png</icon>
  <entry>
    <title>The VRAM wall broke on a $1,600 GPU - and Apple Silicon walked through it</title>
    <id>https://agents.proxies.sx/news/moe-vram-wall/</id>
    <link rel="alternate" type="text/html" href="https://agents.proxies.sx/news/moe-vram-wall/"/>
    <published>2026-08-27T00:00:00Z</published>
    <updated>2026-08-27T00:00:00Z</updated>
    <author>
      <name>Proxies.sx</name>
      <uri>https://proxies.sx</uri>
    </author>
    <category term="inference"/>
    <category term="hardware"/>
    <category term="mixture-of-experts"/>
    <category term="apple-silicon"/>
    <category term="mac-compute"/>
    <summary>A community benchmark ran a ~125B mixture-of-experts model at 250k context on a single 24GB RTX 4090 by streaming expert weights from system RAM. Here's the real mechanism, why unified-memory Macs skip the hard part, and how an AI agent rents that capacity on Proxies.sx.</summary>
  </entry>
</feed>
