Two Minute Papers · 2026-07-01 · notable
Two Minute Papers: 'This New AI Model Changes Everything'
Two Minute Papers walks through GLM-5.2, Z.ai's open-weight coding model with a 1M-token context window, framing it as the first open model to plausibly close the gap on frontier closed labs.
Two Minute Papers frames Z.ai's GLM-5.2 as the open-weight coding model that finally lets you swap out a frontier closed API without giving up much.
What is it?
A Two Minute Papers video on GLM-5.2, Z.ai's open-weight coding-and-agent model. GLM-5.2 ships with a usable 1-million-token context window and, in the developer docs Z.ai points at, is positioned as a drop-in replacement for Anthropic's Sonnet and Opus inside Claude Code.
How does it work?
GLM-5.2 exposes two thinking-effort levels — mapped by Z.ai's docs to 'high' and 'max' — and pairs with a lighter GLM-4.5-air variant for the small-model tier. The video zooms in on how a 1M-token window lets the model plan across a whole repo in one turn, and points at the GLM Coding Plan as the way to actually run it.
Why does it matter?
GLM-5.2 landed with community reactions calling it the step change for open agents, and putting the same model behind a coding CLI at commodity prices lowers the reason to stay locked to a closed API. For teams that ruled out open weights on quality, the video argues that gate has moved.
Who is it for?
Coding agents, indie devs, and teams evaluating open-weights alternatives to Claude and GPT
Try it
https://docs.z.ai/devpack/latest-model