How I AI
How I AI

GLM 5.2: why I’m replacing Opus in Claude Code with this new model

June 24, 2026

AI Summary

5 min read

GLM 5.2, an open-weight model from the Beijing-based startup Z.AI, is being positioned as a serious contender for coding tasks that typically require expensive frontier models like Claude Opus or GPT-5.5. The host tests this claim by running the model through his own stack—Cursor and Claude Code—using OpenRouter as the inference provider. Over the course of a few hours and a few dollars, he finds that GLM 5.2 handles codebase exploration, HTML/CSS design, and long-running autonomous bug triage with surprising competence, though it stumbles on React and TypeScript. The episode is a practical walkthrough of how to set up and evaluate an open-weight model for daily coding work, with honest notes on where it shines and where it falls short.

What GLM 5.2 Is and Why It Matters

GLM stands for "General Language Model," and 5.2 is the latest release from Z.AI. It is an open-weight model, meaning the trained weights are publicly available for download. This allows developers to run it on their own hardware, fine-tune it on proprietary data, and inspect its internals. The license varies, so "open weight" does not automatically mean free for commercial use, but it does mean you are not locked into a single vendor's API. You can self-host or use a provider like OpenRouter, which is what the host does here.

Continue reading the full summary in the app — free to try.

Read Full Summary →

Free • No credit card required

Listen to Audio Summary Open in App

What you'll learn

  • 1 (00:00) **Introduction & Thesis** - Claims that GLM 5.2 delivers Opus-level reasoning at a fraction of the cost, kicking off a review of open-weight models vs. commercial providers.
  • 2 (01:40) **What GLM 5.2 Is (And Why It Matters)** - Explains the model's origin (Z.AI, Beijing), its open-weight nature, and the key benefits: self-hosting, fine-tuning, and vendor independence.
  • 3 (03:10) **Benchmarks & Capabilities** - Reviews the model's specs and third-party benchmark performance against Opus, GPT 5.5, and Gemini.
  • 4 (06:04) **Setup Guide: Getting GLM 5.2 Into Your Coding Stack** - Step-by-step instructions for configuring OpenRouter, then connecting to Cursor and Claude Code.
  • 5 (11:04) **Test 1: Codebase Exploration** - Tests the model's ability to autonomously audit and summarize an unfamiliar codebase (Chat PRD).
  • 6 (12:44) **Test 2: Visual Architecture Overview (HTML)** - Asks the model to create an HTML page communicating the app's architecture and roadmap.
  • 7 (16:38) **Test 3: Real Design Task (Header Redesign)** - Directs the model to redesign a specific hero section of the How I AI landing page, matching an existing design system.

+ Full timestamped outline available in the app

Show Notes

I put GLM 5.2, the open-weight coding model from Z.AI, through four real tasks inside my actual codebase: a codebase architecture audit, a UI redesign, and a 45-minute autonomous bug-hunting session pulling from Sentry and Vercel logs. Total cost: $3.36 for roughly 6 million tokens, a prioritized bug-fix dashboard I’m actually shipping from, and a landing page redesign that matched Chat PRD’s design system on the first try.


What you’ll learn:

  1. What “open-weight” actually means and why it matters for cost and vendor independence
  2. How to connect GLM 5.2 to Cursor and Claude Code
  3. How it performs on codebase exploration and autonomous architecture summarization in a real production Next.js app
  4. Whether GLM 5.2 can match an existing design system
  5. How the model handles a 45-minute long-running autonomous task
  6. Where GLM 5.2 stumbled 
  7. The actual cost breakdown

—

Brought to you by:

Mercury—Radically different banking loved by over 300K entrepreneurs

—

In this episode, we cover:

(00:00) What open-weight models are and why GLM 5.2 is worth testing

(01:38) GLM 5.2 model overview

(04:02) Capabilities and benchmark results

(06:02) How to set up GLM 5.2 in Cursor

(08:37) How to set up GLM 5.2 in Claude Code

(11:04) Live test 1: codebase exploration and architecture audit on ChatPRD

(12:43) Live test 2: generating an HTML architecture and roadmap page

(16:37) Live test 3: redesigning the How I AI landing page in Cursor

(20:57) Live test 4: 45-minute autonomous task, pulling Sentry errors and Vercel logs

(22:35) Where it struggled

(23:49) My verdict on the output

(25:23) Cost breakdown

—

Tools referenced:

—

Other references:

How I AI

More from this podcast

 How I AI →