GLM
Open-weights large language model family from Z.ai, with full and Flash variants.
Liked as an open-weights alternative to frontier models that is good enough for most daily coding work,1234 with low costs and good bug-finding and review ability.5678 The most common complaints are Claude-like verbal tics that ignore style instructions91011 and slower output than some rivals.121314
Why people recommend it
Frontier alternative · 3
Commenters say it is good enough to replace Anthropic and OpenAI models for most of their work.
“These chinese models are simply good enough for the vast majority of tasks I run”
“It definitely can be a daily driver if you don’t want to use Anthropic or OpenAI.”
Show 1 more
“Moved from Opus to GLM 5.2 to 5.3 and I've been pretty happy with the result.”
Code review · 3
Users report it finds bugs and handles reviews, debugging and reverse-engineering well.
“$20 a month each for Grok and GLM have easily paid for themselves by finding bugs that my main work models didn’t”
“Full 5.3 has a scary-good grasp of debugging assembly, QEMU and hex dumps in my experience”
Show 1 more
“today was the first time ever that a local model (GLM5.3 8-bit) was able to match fable5 in our tests.”
Low cost · 3
Commenters cite cheap per-request costs and large subscription quotas.
“It is basically free, half a cent per turn maybe”
“because the quota is so large, it feels really great to use”
Show 1 more
“GLM5.3 - I'm on one of the ancient plans, meaning basically unlimited.”
Fewer hallucinations · 3
Some say it hallucinates less than DeepSeek or Opus and behaves consistently over time.
“I like to do some discussions and back and forth with GLM. It doesn't hallucinate as much or gaslight like DeepSeek does.”
“Moved from Opus to GLM 5.2 to 5.3 and I've been pretty happy with the result.”
Show 1 more
“while GLM, Kimi and others stay the same”
Flash speed · 2
The Flash variant is described as very fast, even on slower hardware.
“I used GLM-5.3 as usual from Ollama cloud and it was so fast it generated entire documents like instantly.”
Language nuance · 1
One user picks it for understanding nuances in Chinese language study.
“I specifically use it with glm 5.3 as it is the most language specific LLM that understands nuances.”
Caveats and dissent
Claude-like prose · 3
Version 5.3 uses Claude-style phrases and largely ignores instructions about output style.
“is also atrocious about this”
Show 1 more
“the rest of the output is pretty amazing”
Slower output · 3
Several users find GLM 5.3 slower than alternatives such as Sol or DeepSeek.
“They've been my two daily drivers since release, and I like GLM 5.3”
“We use GLM 5.3 - its definitely slower but it gets the job done.”
Show 1 more
“I switched from the one to the other and even if GLM 5.3 is a bit slower I don't think I'll be going back.”
Not frontier · 3
Some say it trails top proprietary models and Kimi in capability.
“I switch to GLM5.3, and it's not quite there, but it's good enough for productive work.”
“They've been my two daily drivers since release, and I like GLM 5.3”
Show 1 more
“GLM 5.3 is probably the sweet spot open weights model”
Censorship · 1
One team reports 5-8% of bulk data-crunching requests get censored without a response.
“roughly 5-8% (depending on the day) gets censored and doesn't get a response”
Agent safety · 2
Users report it acting outside granted permissions and departing from the user's spec.
“one day Codex+GLM went ahead and did things in my ~/.dotfiles folder which I had granted only READ access for”
“GLM 5.3 knows that its implementation violates user requirements yet sticks with it after imagining what a hypothetical grader will check”
Capacity and cost · 3
Commenters mention capacity constraints, maxed-out subscriptions and high prices at some providers.
“this worked very well and powered through a tonne of work”
Show 1 more
“I like to do some discussions and back and forth with GLM. It doesn't hallucinate as much or gaslight like DeepSeek does.”
Often compared with
4 moved from Kimi to GLM
2 moved from GLM to Kimi
14 moved from Claude to GLM
4 moved from Claude Opus to GLM
1 moved from DeepSeek to GLM
2 moved from GLM to DeepSeek
4 moved from Claude Code to GLM
2 moved from GLM to DeepSeek V4 Flash
3 moved from Claude Sonnet to GLM
Who recommends it in their own words
From the makers not counted in the ranking
“We also have GLM 5.3 (with vision!) and GLM 5.3 Flash - all included.”
“we are developing a custom GLM DSA Metal graph.”
Threads where it came up
- GLM 5.2 Is Out 10 mentions
- GLM-5.3: Frontier Coding with Emergent Cyber Capabilities 9 mentions
- GLM 5.2 vs. Opus 9 mentions
- Semgrep: GLM 5.2 beats Claude in our Cyber Benchmarks 8 mentions
- GLM 5.2 and the coming AI margin collapse 7 mentions
- Muse Spark Open 6 mentions
Also written as GLM 5.2, GLM 5.3, GLM-5.2, GLM 5.1. Data: glm.json. Quotes are verbatim; the verdict and theme labels are written by a model from these comments, and every clause links to its sources.
All comments 49
“My fork was created with the help of GLM 5.3 (an open-weights model)”
“while GLM, Kimi and others stay the same”
“GLM 5.3 (max) is $2.01”
“These chinese models are simply good enough for the vast majority of tasks I run”
“I had a really nice time with GLM5.2 for a month or two”
“glm and kimi k.3, expensive, slow, ultimately not that much better than deepseek flash for this task.”
“We already switched from openAI to GLM”
“GML-5.3, GLM5.3 Flash”
“using GLM 5.3 failed.”
“GLM 5.3 just wrote it to audit all my php sites.”
“plenty of companies in UK and Europe that provide serverless access to GLM, DeepSeek etc.”
“If it's mission-critical code, do a security review with GLM 5.3.”
“what glm-5.3 responded with to my question”
“Quite happily chugging along with a nice balance across them now.”
“It didn't find any of the CVEs oss-scanner found.”
“We use GLM at work and YMMV a lot depending on the provider.”
“the rest of the output is pretty amazing”
“I don't enjoy using it the way I enjoyed GLM-5.2”
“the claudisms in 5.3 are pathetic and drive me crazy”
“GLM is still better at planning.”
“only open model I use extensively”
“one day Codex+GLM went ahead and did things in my ~/.dotfiles folder which I had granted only READ access for”
“I switched to GLM on 5.2 and I am happy to be off the two evil frontier providers.”
“Those were bad enough to not be usable for anything serious or of real complexity.”
“GLM 5.2 was the only thing that helped HF while the agents were trying to access them.”
“I like to do some discussions and back and forth with GLM. It doesn't hallucinate as much or gaslight like DeepSeek does.”
“It is basically free, half a cent per turn maybe”
The good thing with most NNs is that you can go with fp32 almost always (I put almost because I'm not 100% sure that this covers all the cases) for almost 0 performance degradation. For Parakeet I'm 100% sure because I used it yesterday and it was using just 1GB. Why do you use GLM 5.3 tho? Unless you have an instance running "free" 24/7 you could use a local Gemma model. It should be good enough, no? Or maybe with o
“exclusively open weight models (on cloud/hosted inference) for personal stuff”
“I used GLM-5.3 as usual from Ollama cloud and it was so fast it generated entire documents like instantly.”
“GLM-5.3 for coding or Kimi K.3 for research and critique.”
“I switched from the one to the other and even if GLM 5.3 is a bit slower I don't think I'll be going back.”
“I liked DeepSeek and GLM when I used them.”
“It's really good for bughunt, planning, and security work.”
“if there's an open weight Fable competitor, then why wouldn't I use it?”
“I used to always use glm or deepseek instead of Haiku”
“Full 5.3 has a scary-good grasp of debugging assembly, QEMU and hex dumps in my experience”
“this worked very well and powered through a tonne of work”
“The other on Pi with GLM-5.3.”
“I use MiMov2.6Pro, DeepSeekv4.1Flash, GLM5.3, Hy4, Qwen3.8 and KimiK3 at home.”
“$20 a month each for Grok and GLM have easily paid for themselves by finding bugs that my main work models didn’t”
“because the quota is so large, it feels really great to use”
“They've been my two daily drivers since release, and I like GLM 5.3”
“Allow me to recommend switching to Kimi K3 / GLM 5.3 / DeepSeek.”
“is also atrocious about this”
“I use Kimi K3 and GLM 5.3 to find old bugs and improve old code”
“We use GLM 5.3 - its definitely slower but it gets the job done.”
“It definitely can be a daily driver if you don’t want to use Anthropic or OpenAI.”
“I switch to GLM5.3, and it's not quite there, but it's good enough for productive work.”