Seven LLMs running in your browser tab, no server involved
Someone built a lab where you can test tiny language models that run entirely client-side. No API calls, no backend, just WebGPU and a lot of patience.
Read the note 2 min read
From the notebook
edge-computing.
3 notes on this topic.
Someone fit a reasoning model into 14MB
Needle2 runs on phones, smartwatches, and Raspberry Pis. The entire model is smaller than a single photo.
Read the note 1 min read
Cloudflare just put feature flags at the edge
Feature flags that run on Cloudflare's edge network, not your backend. This might actually change how you roll out features.
Read the note 1 min read