‹ BackHN Continuity

Thread

MicroLLM Lab – Try 7 tiny LLM's in the browser

283 points · 113 comments · logicallee

  1. willaaam · · focus · HN ↗
    I like it as I'm vibecoding an (airgappable) browser AI workspace myself, but in terms of putting the models to use, just exposing the chat interface feels a bit limiting to me.

    My take on this concept: <a href="https:&#x2F;&#x2F;github.com&#x2F;willaaam&#x2F;gemma-4-E2B-webgpu-vision" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;willaaam&#x2F;gemma-4-E2B-webgpu-vision

    1. logicallee · · focus · HN ↗
      It&#x27;s a cool project, but I couldn&#x27;t get it to load. What did you test this on? I tried the live link here:

      <a href="https:&#x2F;&#x2F;willaaam.github.io&#x2F;gemma-4-E2B-webgpu-vision&#x2F;" rel="nofollow">https:&#x2F;&#x2F;willaaam.github.io&#x2F;gemma-4-E2B-webgpu-vision&#x2F;

      And after loading it, with an NVidia 1060 GPU (6 GB RAM) on Windows it failed with &quot;Failed to load: No supported WebGPU variant for com.xenova.gemma4.DenseGemv; rejected sgma&quot;.

      In Safari on a 2026 Mac Mini M4 with 24 GB of RAM it failed with &quot;Failed to load: JSON Parse error: Unexpected EOF&quot;.

      The idea is pretty cool though!

      1. willaaam · · focus · HN ↗
        Huh, weird! It doesn&#x27;t work in Firefox, but I test on Safari (M5 Max) and Chrome (Linux, Arc B580). I don&#x27;t have access to NVidia or AMD hardware myself unfortunately, though friends confirmed both as working.

        Going to debug tomorrow!

Open on Hacker News to reply ↗

Unofficial Hacker News client; not affiliated with Y Combinator.