GLM 5.2

The difference is glm is 1/10th of the cost per token and has way higher usage limit

You forgot to mention that GLM-5.2 can also run locally if you have enough hardware like 256GB Mac Studio or a system with sufficient VRAM + RAM. Unsloth also provides a 2-bit quantized version that reportedly retains 82% of the original model accuracy, making local deployment much more accessible.

This was ABSOLUTELY INSANE. Are you choosing GLM 5.2 or Claude to use in 2026? I genuinely can't decide anymore lol 😭😭

Announcement from the founder of Z.ai: “ GLM-5.2 is Fully Open, Frontier Intelligence Belongs to Everyone Today, the sudden restriction of certain frontier models is deeply regrettable. At a time when access to frontier models is abruptly cut off for non-technical reasons, we are even more convinced of one thing: science should be global. The path to AGI (Artificial General Intelligence) must never be enclosed by high walls. We have always believed that AGI should be the cornerstone for all of humanity to collaboratively explore the boundaries of intelligence and solve complex challenges, rather than a privilege monopolized by a few rules and subject to revocation at any moment. In the face of external blockades and restrictions, our attitude is one of radical openness. Frontier intelligence must remain open-source, accessible, and buildable, serving every dedicated developer. GLM-5.2 is Zhipu's most capable open-source model to date. It not only supports a truly usable 1M context window but also maintains a continuous lead in the independent completion of long-horizon tasks, providing solid foundational support for building complex agent applications. It also continues to be our main engine for creating the strongest domestic coding model. Tonight at 5:21—at this special moment—GLM-5.2 will officially be available to all GLM Coding Plan users (including Lite / Pro / Max). The API will also go live next week. A step closer to frontier intelligence for everyone. The future of AI is open, and it is for the people. ModelKey: GLM-5.2” https://x.com/jietang/status/2065784751345287314

NOBODY can make a new AI model by distillation alone, NOBODY. in fact, every model i know of distilled from existing models during the training/verification process! so accusing anyone of doing distillation is at very least disingenuous, if not extremely hypocritical.

I run Q4_K_XL. All it takes to run to get about 6tk/sec is 512gb of ram and 2 3090 GPUs with llama.cpp -cmoe. I also have crappy DDR4, 2400mhz, 3200mhz will bring that speed up to about 9tk/sec. I also have ok 32core epyc CPU, a better 64core would bring it up to about 11tk/sec. I did a budget build before the crazy hardware cost and I regret it everyday. Nevertheless, it's fantastic being able to run this model at home. It's great for planning, one shot prompting once you have a plan or all the context you need. This entire hardware cost $2400 when it was built. If you're willing to be resourceful, you can find ways to run these models at home. I often get the silly question of why, and suggestions about how much I can save using cloud API, but the Fable drama has opened up eyes on why it's good for us to be independent. Thanks team unsloth, Q4_K_XL is solid, if you are going to grab a quant, make sure to get the K_XL variant if it can fit.

I have taken another look on these open models after the fiasco of Fable and GPT 5.6 this weekend and... GLM-5.2 truly is a good workhorse model for daily programming. I consider myself a heavy user of LLMs and a seasoned developer. A typical session for me with GPT is usually over a hundred dollars... This weekend I programmed a matrix bot with encryption and a Rust agent with some tools. Because I need one and OpenClaw just felt... not what I wanted. Two days later and 20 dollars poorer I have what I need: a multimodal agent written in rust that has access to my homelab. Nothing felt off with GLM. It did what I wanted, was fast, had a decent not very annoying personality and was much cheaper than Opus or GPT. I used it unquantized through Fireworks, but there are multiple other providers too.

The Chinese are being responsible humans with open source while the oligarchy Americans are just trying to make more money even if AI ends up killing us all…

>Don't worry though, open source evangelists will tell you that these will be running on your phone in the next 3 years. Not sure if you're being sarcastic, but I can run a quantised version of Gemma or Qwen on my 16GB M1 Macbook Pro that beats GPT-4 from 2023 hands-down. I wouldn't be surprised if, in another 3 years, you'd be able to run something as powerful as Opus 4.5 or GLM-5.2 on standard consumer hardware - say a 32GB/64GB M7 Pro. I also wouldn't be surprised if, 3 years after that, cheaper hardware and improved model efficiency means that there's a much smaller gap between what you can run on a consumer CPU (which, with memory prices coming down, could look like a 256GB M9 or M10 Pro) and $100k GPU cluster.

Glm5.2 es espectacular 😮

A new cow to milk has arrived🥹

GLM5.2 is still not cheap, at 1/5th the price of Opus 4.8 API token pricing, it is still hefty. What we really need is the next Deepseek model to come in at Fable 5 level of performance and cost same as V4 and with similar cache hit levels and cache hit pricing.....now that will be the real game changer and once and for all shut down the evil efforts from Anthropic and OpenAI to get Open Source models banned or made difficult to access !!!

The full GLM-5.2 has 1.4 terabytes of data. Even in quantized form, reducing that size by 2 times or 4 times, a Raspberry Pi 5 could not run it otherwise than by reading the weights from an SSD. Even thus, I do not believe that a Raspberry Pi 5 would be fast enough to be able to run inference on such a big model at the speed at which it can read from the SSD. On the other hand, there are many mini-PCs with Intel or AMD CPUs that have both a PCIe 5.0 SSD and a PCIe 4.0 SSD, which may be read in parallel, achieving thus a reading throughput of up to 20 Gbyte/s. Such miniPCs have fast enough CPUs/GPUs, so that they might be able to reach the inference speed limited by a 20 Gbyte/s weight reading throughput, which for a so big model like GLM-5.2 would be of one output token every few seconds (only a fraction of the weights must be read for one output token). The ratio between output tokens per second and the weight reading throughput can be improved by various methods, like multi-token prediction or batching multiple tasks. Optimizing inference speed in such conditions is an active research subject, due to the high current memory prices.

I agree. And at work it has been producing some of the worst GUI test cases I have ever seen. What is tested often makes no sense at all, completely implausible edge cases are tested on internals, while it doesn't create tests for the overall application using user events. And some things in these test cases are downright ridiculous: instead of instantiating your classes, it sets up some barebones fake objects reimplementing some of the behavior of your actual class, then ignores the TypeScript errors via force cast or similar. Then it proceeds to slap some test ids on the output, stubs components and dependencies more or less randomly, adds some assertions on test ids and calls it a day. Apparently that's good enough for many colleagues to open a MR for that garbage. That said, at home with SOTA models I happily hand large units of work to it, outsource much of the thinking, and get workable results. I think this is the future.

It seems to really be a nice step-up and is getting quite close to the frontier. I wish they'd start focusing on the reasoning efficiency now, though. I have a simple (relatively) test task to evaluate LLMs: writing a simple math evaluator library in Nim (it's about 400-600 lines total max), and GLM 5.2 (xhigh which maps to max effort) spent over 15 minutes (!) reasoning, spending about 45k tokens, before it finally wrote the first file. I know it's hard to improve on that, but now that their models are good enough at raw intelligence, I think this should become a higher priority task. Currently on https://artificialanalysis.ai/#output-tokens GPT 5.5 xhigh spends 16k tokens total on average, GPT 5.5 high is 10k, Fable 5 33k, Opus 4.8 41k, GLM 5.2 is 42k. GPT 5.5 is extremely reasoning efficient. Of course if you convert those values to actual request cost, GLM 5.2 will probably beat GPT 5.5/Opus 4.8, but speed matters for a lot of people, I think.

Agree'd here. I am seeing a pattern in this thread - Everyone here seems to justify paying substantially more money to self-host hardware thinking they are saving in the medium-long term. Frankly from the perspective of the company I'm with - lost opportunity cost is more important than trying to future proof on a hunch that AI vendors are going to pull the board out under you. There's alot of competition in this space and we're seeing these frontier models continue to get better for the foreseeable future. As such, you're dumping alot of money upfront on hardware which is not guaranteed to be relevant in a few years time. The benefits to me seem pretty speculative, minus the point of security.

My anecdotal experience differs (though I hold ground that LLM evaluations are highly subjective and benchmarks are just as useful for LLMs as they are for dating websites users). GLM 5.2 tends to stray way more than and 5.1. It also hallucinates you things subtly: morphs requirements, makes unfounded conclusions. This output is not something I experienced in any model I seen so far. In coding it's especially annoying because it steers whole request. E.g. I give instruction: "make we a Rust-WASM-Canvas app" and GLM 5.2 goes like "Oh user surely doesn't mean that. I'll better build Dioxus app instead".

Based on DeepSWE, Opus 4.8 gets you more intelligent output at lower price (GLM's token inefficiency is really biting them). GPT5.5 even moreso. And I don't recall about Opus but GPT is much, much faster at getting you the answer (again, GLM's token inefficiency). It's neat, I guess, that we can compare them against models released last year, but I care about my options today, and the pareto frontier is about as far away as it ever was. Add on top of that the extra features OpenAI and Anthropic have in their apps and...

That's right, but there are other recent open weights and relatively big LLMs that are multimodal, e.g. MiniMax-M3. With open weights LLMs, it is affordable to use many different models, each for whatever it is better. Moreover, for analyzing "UIs, photos, screenshots, etc." there are small models that can be run locally on smartphones or laptops, e.g. IBM granite-vision-4.1-4B, certain Google Gemma 4 variants and certain Qwen variants, whose output you can use as input for a big LLM, in order to accomplish some more complex task.

6:50 AI slop company invents a compiler