GLM5

FIRST-HAND READS

What people reported, in their own words. Every line links to where it was said.

• “GLM5.2 is, in my view, a marvel; it's inexpensive and works very well when used properly”

  POSITIVE · performance · Hacker News · 2026-07-10

• “GLM5.2 is too jumpy and eager to do things, often to a fault (e.g. deploying/using git when it shouldn't)”

  NEGATIVE · performance · Hacker News · 2026-06-30

• “GLM5.2 is also excellent, it found things in our codebase that 5.5 and opus 4.8 both missed.”

  POSITIVE · performance · Hacker News · 2026-06-27

• “I just let GLM5.2 run because I was convinced everything was good now, but it also found two good bugs that Fable missed.”

  POSITIVE · performance · Hacker News · 2026-06-20

• “GLM5.2 ends up being far more expensive than I thought it would be”

  NEGATIVE · price · Hacker News · 2026-06-17

• “glm5.1 is far more cost effective for us than Opus with the rate of defects and rework taken into account”

  POSITIVE · price · Hacker News · 2026-05-28

• “GLM5 is at this point much more capable than Claude and Codex for complex backend end work, complex feature planning, and long horizon tasks”

  MIXED · features · Hacker News · 2026-04-08

• “Al the others failed but glm5.1 found a set of parameters and tweaks how to make my browser in container not be detected.”

  POSITIVE · performance · Hacker News · 2026-04-07

• “glm5.1 is producing mich better typescript than opus or codex imo”

  MIXED · performance · Hacker News · 2026-04-07

• “glm5 was broken vs another provider, returning gibberish and very dumb on the same prompt”

  NEGATIVE · performance · Hacker News · 2026-03-21

• “how bad GLM5 has gotten with complete radio silence from zai team. What started as groundbreaking open source model and affordable coding plan was heavily quanted / KV cache quanted ~14 days ago leading to model bleeding thinking tokens into responses, no assistant messages, repetition loops at very low context sizes (despite claimed ~200k context)”

  NEGATIVE · performance · Hacker News · 2026-03-15

• “The actual writing code is GLM5 - once properly spec'd out it can generally just hammer away at it till its done”

  POSITIVE · performance · Hacker News · 2026-03-14

• “GLM5, Gemini 3 Pro Preview, and Claude Opus 4.6 all correctly identified the problem and said Drive.”

  POSITIVE · performance · Hacker News · 2026-02-16

• “stuff with recent models (gemini 3, claude 4.5/6, even smaller, open models like GLM5 and Qwen3-coder-next) that was just unthinkable a few months back”

  POSITIVE · performance · Hacker News · 2026-02-14