<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0">
  <channel>
    <title>The Hitchhiker's Guide to Open-Weight Models — Changelog</title>
    <link>https://anotheremoji.com</link>
    <description>Weekly updates to the Hitchhiker's Guide to Open-Weight Models. New models added, models considered and dropped, and the occasional towel advisory.</description>
    <language>en</language>
    <lastBuild>Mon, 12 Oct 2026 09:15:00 +0100</lastBuild>
    <item>
      <title>October 12, 2026 — 4 ADDED, 1 UPDATED (backfill: Gemma 4, Command A+, North Mini Code, Qwen3.8-Flash-Next)</title>
      <link>https://anotheremoji.com/changelog.html</link>
      <guid>https://anotheremoji.com/changelog.html#20261012</guid>
      <pubDate>Mon, 12 Oct 2026 09:00:00 +0100</pubDate>
      <description>A quiet release week spent checking the Guide&#39;s own homework: four significant models that shipped between April and August escaped the initial survey and are now aboard — Google&#39;s Gemma 4 (the biggest miss), Cohere&#39;s Apache-2.0 flagship Command A+, the 30B-A3B North Mini Code edge coder, and Qwen3.8-Flash-Next, the 180B-A6B preview of the Qwen4 architecture now anchoring Hugging Face trending. Mistral&#39;s 1T Le Chonk waits for its October 27 weight drop.</description>
    </item>
    <item>
      <title>October 5, 2026 — 1 ADDED (Aleph Alpha Kolibri, Ling 3.1 Flash +4 more considered)</title>
      <link>https://anotheremoji.com/changelog.html</link>
      <guid>https://anotheremoji.com/changelog.html#20261005</guid>
      <pubDate>Mon, 05 Oct 2026 09:00:00 +0100</pubDate>
      <description>Europe finally ships: Aleph Alpha's Kolibri (78B/3.46B German-English MoE, Apache 2.0) lands on Reunification Day as the Guide's first European entry. Meanwhile the frontier stayed quiet — Ling 3.1 Flash is API-only, Cloudflare's Clef is a classifier wearing a trench coat, and Nebius Reflection is still a rumour with a press release.</description>
    </item>
    <item>
      <title>September 28, 2026 — 4 ADDED (MiMo-V2.6-Pro, MiMo-V2.6-Flash, GLM-5.3-Flash +4 more)</title>
      <link>https://anotheremoji.com/changelog.html</link>
      <guid>https://anotheremoji.com/changelog.html#20260928</guid>
      <pubDate>Mon, 28 Sep 2026 09:00:00 +0100</pubDate>
      <description>The week the open-weight crown changed hands. Xiaomi — purveyor of phones, rice cookers, and electric cars — shipped MiMo-V2.6 under MIT, and the Pro variant immediately became the highest-scoring open-weight model ever measured (AA Index 46.32). Plus a GLM-5.3-Flash backfill and a 5.9 GB ternary model that ends the 2-bit excuse...</description>
    </item>
    <item>
      <title>September 21, 2026 — 1 ADDED (Atria Dawn Preview, Meta Muse Spark 1.3, Nex-N2.5 Max +3 more)</title>
      <link>https://anotheremoji.com/changelog.html</link>
      <guid>https://anotheremoji.com/changelog.html#20260921</guid>
      <pubDate>Mon, 21 Sep 2026 09:00:00 +0100</pubDate>
      <description>A quiet week for open-weight releases, but one entry cleared the bar: Shanghai AI Lab quietly posted a 744B agentic MoE on Hugging Face without a launch event, then published a 143-author paper three days later. The model beats Claude Opus 5 on five agent benchmarks. Meanwhile, Meta shipped Muse Spa...</description>
    </item>
    <item>
      <title>September 14, 2026 — 2 ADDED (IBM Granite 4.2 (30B), MiniCPM5-2B, Inception Mercury 2.5 +2 more)</title>
      <link>https://anotheremoji.com/changelog.html</link>
      <guid>https://anotheremoji.com/changelog.html#20260914</guid>
      <pubDate>Mon, 14 Sep 2026 09:00:00 +0100</pubDate>
      <description>A quieter week at the frontier — the big labs shipped GPT-6 Astra, Claude Fable 5.1, and Gemini 3.8 Flash, but all proprietary. The open-weight action was in the small and edge tiers, where IBM finally shipped a reasoning model worth downloading and OpenBMB proved that 2.5 billion parameters is enou...</description>
    </item>
    <item>
      <title>September 12, 2026 (Server Sizing Guide)</title>
      <link>https://anotheremoji.com/changelog.html</link>
      <guid>https://anotheremoji.com/changelog.html#20260912</guid>
      <pubDate>Sat, 12 Sep 2026 09:00:00 +0100</pubDate>
      <description>A new page has appeared in the Guide. Not a model — something arguably more useful: a practical guide to not running out of VRAM. Because the most common sound in the known universe, after &quot;hello&quot;, is &quot;CUDA out of memory.&quot;</description>
    </item>
    <item>
      <title>September 12, 2026 — 4 ADDED, 2 UPDATED (DeepSeek V4.1-Flash, Tencent Hy4 Preview, K2 Horizon (375B-A23B) +6 more)</title>
      <link>https://anotheremoji.com/changelog.html</link>
      <guid>https://anotheremoji.com/changelog.html#20260912</guid>
      <pubDate>Sat, 12 Sep 2026 09:00:00 +0100</pubDate>
      <description>Week two of the open-weight galaxy survey, and it was a busy one. DeepSeek replaced its own flagship with a cheaper model that beats it — a move that either redefines the frontier or is a spectacular own goal, depending on your perspective. Tencent dropped 770B under Apache 2.0, IFM released the mos...</description>
    </item>
    <item>
      <title>August 31, 2026 — 4 ADDED (GLM-5.3, Qwen3.8-2.4T-A95B, Qwen3.8-27B +5 more)</title>
      <link>https://anotheremoji.com/changelog.html</link>
      <guid>https://anotheremoji.com/changelog.html#20260831</guid>
      <pubDate>Mon, 31 Aug 2026 09:00:00 +0100</pubDate>
      <description>First weekly scan of the open-weight galaxy. Four significant new models detected and added to the Guide. The galaxy got noticeably more crowded this week — Meta reversed its closed-weights pivot, Z.AI held weights for two weeks over cyber capability concerns, and Alibaba decided &quot;Max&quot; models should...</description>
    </item>
  </channel>
</rss>
