GLM 5.2 Clearly Explained (and how to set it up)

In this episode I sit down with Amir to get tactical about running local AI models as part of a daily workflow. We center on GLM 5.2 from ZAI, how it stacks up against frontier models like Opus 4.8, and how a fusion approach lets you sequence a heavy thinking model with a lighter execution model for the best output at the lowest cost. Amir walks through setup in Cursor and Codex via OpenRouter, shares real token-cost math, and demos GLM 5.2 refining a live app. By the end you will know how to start today, where local models shine, and how model chaining keeps spend in check. Timestamps 00:00 – Intro 02:09 – GLM 5.2 and Z AI 04:01 – Specs: 1M context and Terminal Bench 2.1 05:22 – Making sense of benchmark scores 06:42 – Setup in Cursor or Codex with OpenRouter 10:18 – Local model upside: buy a machine, run tasks 11:42 – Token cost: 44 cents versus $2.38 13:36 – Future-proofing with an upfront hardware bet & The Uber subsidy analogy 16:49 – Model chaining and the vision workaround 19:23 – Token maxing vs routing tasks to the right model 20:54 – Answering the "cost is irrelevant" crowd 21:59 – Closing thoughts Key Points GLM 5.2 ships with a 1M-token context window and scores 81 on Terminal Bench 2.1, landing about four points behind Opus 4.8. A fusion approach (a term OpenRouter coined) sequences models: plan with Opus, execute with GLM 5.2, review with Composer 2.5 or Codex 5.5. Running GLM 5.2 in the cloud through OpenRouter costs roughly 44 cents for a task that runs about $2.38 on Opus 4.8 — close to a 5X saving. You can start today with credit-based access: load $20 in OpenRouter and route tasks to the right model. For images, Amir uses Opus 4.8 to read screenshots and describe them, then hands the layout to GLM 5.2 to act on. Teams are shifting from token-maxing to output-maxing, making model governance and chaining the smart play The #1 tool to find startup ideas/trends - https://www.ideabrowser.com LCA helps Fortune 500s and fast-growing startups build their future - from Warner Music to Fortnite to Dropbox. We turn 'what if' into reality with AI, apps, and next-gen products https://latecheckout.agency/ The Vibe Marketer - Resources for people into vibe marketing/marketing with AI: https://www.thevibemarketer.com/ FIND ME ON SOCIAL X/Twitter: https://twitter.com/gregisenberg Instagram: https://instagram.com/gregisenberg/ LinkedIn: https://www.linkedin.com/in/gisenberg/ FIND AMIR ON SOCIAL Humblytics: https://humblytics.com/?via=community X/Twitter: https://x.com/amirmxt Youtube: https://www.youtube.com/@amirmxt

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:00Greg Isenberg?

  Okay, you've probably heard of GLM 5.2 that's going viral everywhere on Twitter. Yes, it's this new open source local AI model that people are saying is the chat GPT moment for local AI. But no one's actually gone and shown you how to use it and how do you actually set it up? So I figured I'd bring on my friend Amir. He tells us exactly how you should think about running GLM 5.2, how you should think about running local models, how that integrates to something called open router, how you can use it with your codecs or cursor or Claude code. And this episode in 20 minutes or less, you're going to get everything you need to know

• 0:39Greg Isenberg?

  about local AI models, why GLM 5.2 is crushing benchmarks and how you can set it up today so you can go and build your startup and build your business and be more productive and be more efficient. Enjoy the episode and I'll see you at the end of it. Hit a like, comment and subscribe for more of this sort of stuff in your feed. Enjoy.

• 1:09Greg Isenberg?

  Welcome to the show, Amir. By the end of this episode, what are we going to learn?

• 1:15Speaker 01

  We're going to talk about essentially learn about how local models are kind of keeping up now with the pace of these closed models as well and how you can kind of use compounding models or fusion models, as open router calls it, to be able to do sequencing between a more extensive thinking model and a more execution based model. We'll show you how GLM 5.2 actually compares and stacks up against other models and how you can effectively use it and get set up with it as well.

• 1:41Greg Isenberg?

  Cool. I did an episode on local models. It was a hit. People wanted more tactical. How do I actually, you know, how should I think about local models? How do I implement local models? How I can actually make this a part of my daily workflow? So I brought on Amir. Welcome to the show and let's, I give you two welcomes, by the way. That's how excited I am to have you share with everyone, everything. So let's get into it.

• 2:07Speaker 01

  I'm super excited to be here. Let's jump right into it. So let's talk about what happened this week. ZAI came out with GLM 5.2. And I think this was a big inflection point because we've typically seen with local models, it's either, you know, storage extensive that you can't essentially run it or install on your computer, or you need a better like GPU RAM performance to be able to actually run it locally as well. Now, GLM 5.2 is also resource extensive, but what we're seeing is with open source models like, uh, open source providers like, um, OpenRouter or Llama, being able to help you run these models in the cloud and effectively being able to essentially pay, slightly less for input and output tokens compared to the more closed models.

• 2:52Speaker 01

  Now, what I want people to take away from this session is, one, how to actually get set up with it. We're not going to go through the detailed setup process, but I'm going to just cover how you can do it in Cursor using OpenRouter or in Codex, and then effectively talk about how GLM 5.2 stacks up against the other models, and then how you can effectively use it as well to do model training. And then we'll do, um, model training, and then we'll do just a kind of a, maybe a quick walkthrough of how I'm currently using it. You know, I want to be very honest, like, um, these, um, local models still have a lot of work to do in terms of, having tool capabilities to be able to, you know, have, um, the mod, you know, the modalities to be able to see images and conceptualize on what they're looking at. And I'm going to tell you how you can effectively, uh, circumvent some of that, where you can use other models to

• 3:37Speaker 01

  explain what the, the image is back to GLM and then have GLM work on it. And then also just have a very live test on how this stacks up against other models. You know, benchmarks are great. Personally, I'm not an expert in it. I don't know what any of these benchmarks actually mean. The way I do is off labs, like, let's build it out and see how this actually looks and how it stacks up against other models. Sound good?

• 4:00Greg Isenberg?

  Yes, sir.

• 4:01Speaker 01

  Let's do it. Okay, so GLM 5.2 came out and essentially it has a 1 million contacts window. And it scores 81 points on the terminal bench 2.1. It's just about four points behind OPUS 4.8. And it does quite well on the long horizon task evaluation. So this is essentially projects you have that you want to run long sequence tasks on. And, you know, I think it's taking into account like the thinking parameters and how it can think through and plan through some of the the task it has ahead. So you can see that across all these different kind of benchmark reporting or reports. GLM 5.2 actually does quite well. So in this case, it's 62.1% compared to OPUS 69.2.

• 4:47Speaker 01

  What's special about GLM is it's open source. You can run it locally on a machine if your device can support it, or you can run it in the cloud through the open model providers. It's a big leap from 5.1. I personally didn't test 5.1. I got straight into 5.2. But from what I'm seeing based off of like Twitter and conversations with people, it's performing quite well, especially on the front end side of just execution based tasks. I haven't really tested more on the backend resource extensive tasks. But just based on perception and what we're seeing in the reporting, it's stacking quite well.

• 5:22Greg Isenberg?

  So when you say stacking well, are we talking like, because like I look at benchmarks and honestly it goes through, you know, one ear, one eye and it goes at the other. I'm like, I glaze over it because I'm like, what does this really mean? Like, are we talking, is it like 4.8? Is it like 5.5?

• 5:38Speaker 01

  Yeah.

• 5:39Greg Isenberg?

  What do we, how should I think about this? And yeah, give it, give it to me straight.

• 5:45Speaker 01

  Honestly, man, I'm gonna say, man, I don't get it. You know what I mean? Like, I'm, I'm not smart enough to understand how these benchmarks actually stack. So I'll be honest, for me, it's like, let's just, let's just build it, use it and see how we feel about, you know, it, how it performs compared to the other models. And for me, I want to get the best out of it. So if I feel like GLM 5.2 is strong in one part, but weak on the other, then I'm thinking about how do I actually use other tools? Or other models to essentially now I think like almost like a fusion approach now I think open router

• 6:17Speaker 01

  which is one of like the model providers coined this where it's like you're able to do like sequencing between two different models to get the best output. So I'm totally game on if I can run a local model on my machine to do certain tasks but then call you know Opus or Codex to do something else and have them work together by all means I want to be the most token and cost efficient and performance as well.

• 6:41Greg Isenberg?

  Yeah.

• 6:42Speaker 01

  So on the setup side, I personally started using this through cursor using open routers API. So how it works essentially is you gotta go to ZAI which is the GLM provider they created the GLM five point two model. You get an API key from them and then you take that key and you go into your cursor settings. Paste it into the open AI field and then from there you override the open AI endpoint with this API endpoint right here. And essentially from there you go back to models add a custom model GLM five point two and you're able to then actually call GLM five point two directly.

• 7:22Speaker 01

  So in essence, instead of open AI key, you put your GLM API key from Z AI. Then you override the API endpoint for when you call open AI chat completion with this one right here. And then you go back into custom models and add the custom model protocol. You can also alternatively do this using open router. So if you want, if you're using Codex, you can go to open router, get your open router key. And then go into the provider, get the endpoint and then go into Codex, create a profile and say, hey, I want you to install this model. Open source model. Codex does actually support open source models. So you're able to provide the details of what the model is, the context window.

• 8:04Speaker 01

  And then essentially when you're running Codex to the CLI, you can switch to GLM five point two.

• 8:11Greg Isenberg?

  Easy enough.

• 8:12Speaker 01

  Yeah, easy enough, you know, and then maybe we can have a page or something to show later on where they can kind of follow these instructions. There's a lot of guides on Twitter and online, you can follow those, but essentially, in my opinion, the best way to get started with this is just go to open router and cursor and get that set up through and through. So let's talk about here the model, we've talked about the benchmarks, how it performs. Really, if we want to just at least takes, you know, put some weight to the benchmarks. I'd say if they're scoring at 62% and OPUS 4.8 is at 69, you know, that probably means something, you know, for for the normal people, we probably won't really know until we actually play around with it. But I went in and was just looking at, for example, this website we have, this is a small app.

• 8:56Speaker 01

  And I built this in, I think OPUS 4.8, I was just testing it around. And I started refining the design using GLM 5.2. So I was like, hey, redesign the hero section for me or refine it. There was this like section right here with all these images. I was like, why don't we just do a little like carousel style? So it, you know, it's fascinating in a way because I don't, I personally don't think open, the, the local models had

• 9:24Speaker 01

  this kind of capability to be able to get, get it so refined and accurate previously. And I, I tested the models like that we had. And I find that GLM 5.2 is a lot more refined and it's able to follow the instructions on like what you want it to do. So in a couple prompts, I was like, hey, you know, let's do a carousel here. Let's make sure we, you know, are able to show the, the images. And then from there, I want you to build out like a bentogrid style of all the features that we have. Now, this is all one prompt, obviously, you know, you can see it's a little bit vibe coded here. It has a little side like badge, the labels, you know, and you can tell. But at the end of the day, for a local model, for like, you know, a local model, if you're running this in a computer and not burning any tokens,

• 10:10Speaker 01

  it's doing, it's doing quite well. And I think I can see how it starts like internally reporting.

• 10:16Greg Isenberg?

  Yeah, I mean, I think so, like, okay, what is the, what is the? The main benefit of using a local model versus something in the cloud is you don't burn tokens. You essentially, the way to think about it is you're buying a machine. And, and correct me if I'm wrong, but you're buying a machine. And we should talk about like some of the machines that people could potentially buy. But you're buying a machine, it's a, you know, one size, it's a one, it's a cost, it could be 2,000, 5,000, 10,000. And then you can just run tasks, right? So you're building a startup, you just want, maybe you want conversion rate optimization.

• 10:58Greg Isenberg?

  So maybe you say every day, I'm going to feed you customer feedback and every single day I want you to work on the front end. And you just do that. So my question to you is for local models specifically, for people who actually want to be building companies, shouldn't they be basically running it all the time on certain tasks? And how should people be thinking about it?

• 11:21Speaker 01

  Yeah, so I think that's a tough, that's a little bit tough to answer and I'll say why. Because like this model specifically is really resource intensive. So a lot of, I think, existing consumer computers may not be able to run this from what I've seen.

• 11:36Greg Isenberg?

  Yes.

• 11:36Speaker 01

  I've been, I've been running it on the cloud through Open Router directly.

• 11:42Greg Isenberg?

  And what's the cost with that?

• 11:44Speaker 01

  Yeah, so with, so I actually was trying to map out the token cost of model training. So if we had about 50,000 input tokens and 85,000 output tokens, to get almost close to an Opus 4.8 level of output, it will cost us 44 cents. Whereas with Opus 4.8, it costs you $2.38.

• 12:08Speaker 01

  So the, you know, there's a big, almost like a big difference on the, almost like 5x, you know, price difference between.

• 12:17Greg Isenberg?

  Which doesn't sound like a lot when you're like, oh, $2 here, 44 cents here. But when you're actually using these things and you're running it all the time and you have pretty, you know, big tasks that you're going after and you don't want to be constrained by token costs. So it's a big deal. 5X is a big deal.

• 12:35Speaker 01

  Yeah, yeah. And this is based on kind of the averages on like the coding benchmarks and what what cursor charges you to the API pool. But I want to, I want to, I want to be also feature proofing ourself. If we got this far with GL 5.2, I wonder what the next six months looks like, right? So would it be even worth, like, I think it's also worth thinking about making an upfront investment in your machine right now to be able to potentially download and run these local models? So that when we get to GL GL 5.3 or 5.5, we're essentially made that upfront investment in the in the in the compute and the equipment to now save a lot more in the long run with other feature models that are going to come out that are going to be much more extensive. Because I think this, you know, we're seeing the the AI subsidy on tokens, right?

• 13:20Speaker 01

  Like, we're getting a lot more output out of Claude, out of Codex. And I wonder, you know, I I've personally seen it, I'm sure you have too, where it's like now we're hitting our usage a lot faster than we before, especially when Fable came out. I ran, I remember, like, I ran it in, like, in the first day I hit my limit, you know?

• 13:35Greg Isenberg?

  Totally. So you're, what you're saying is basically like, you know, if you look at,

• 13:42Greg Isenberg?

  I mean, if you look at the history of VC-backed startups, think about Uber. When Uber first came out, they actually subsidized rides and they got you hooked on to the, on to the, on to the app. And then over time, they started increasing prices, increasing prices. What you're saying is with, you know, in the AI age, with a lot of these LLMs, they're going to get you hooked into the workflows, you're gonna, you're gonna build on top of it. And over time, you know, those subsidies are going to go away as they go public and things like that. So what you're saying is maybe it's a good idea to actually invest in running this thing locally. Because, you know, the price of memory isn't getting cheaper.

• 14:26Greg Isenberg?

  And the price of tokens aren't getting cheaper. So building it now, securing it while you can, might be a good idea.

• 14:34Speaker 01

  Yeah, and to tie this all together, I'd say two things. One, harnesses that are agnostic on models. So like, for example, Cursor, where you're able to run multiple models across the same sequence of tasks, are going to actually potentially benefit from this, right? So, you know, I wouldn't be surprised if one, you know, Cursor decides to directly support Gion 5.2 as a model provider and lets you kind of tap into that cost saving. If you couple it with like Compose 2.5. So this is where it goes into model training, right? Where I was, for example, looking at, earlier in this task here, I wanted to refine the hero section.

• 15:13Speaker 01

  So what I did is I actually used Opus 4.8 to, to first import screenshots, because Gion 5.2 doesn't support vision capabilities. So what I did is I actually used Opus 4.8 to import screenshots and explain back to me what it sees, right? I was like, tell me what you see specifically on the front end design for the hero section and lay it out. And then I switched to Gion 5.2 to steady that layout and then actually act on making those changes. So it's kind of a way to like circumvent the fact that you have limitations to what Gion 5.2 can do on, like image capabilities. But you're able to kind of now train the expensive model to think through the plan

• 15:58Speaker 01

  and then get the same level of frontier, like frontier level, like quality, but at a much affordable price point.

• 16:09Greg Isenberg?

  I mean, that makes sense. So your, your recommendation is basically, you know, there's, it's almost like free trade versus protectionism. You know, the world, you, you, you know, not to be politic, this isn't, this isn't political. This is just an economic theory, right? Which is like, you know, when, when, when people are trading with amongst each other, you know, maybe it's, you know, in Canada where you are, you know, you might want to trade with Florida because, you know, we got good oranges here. And we might want to trade, you know, we, we don't get, we can't make maple syrup here. So we'll get your maple syrup, right?

• 16:45Speaker 01

  We'll get the maple syrup, yeah, exactly. So make the best use of it, yeah.

• 16:49Greg Isenberg?

  Make the best use of it. You're saying is, you know, using Cursor.

• 16:55Greg Isenberg?

  And basically, you don't have to use Cursor, right? You can use whatever you'd like.

• 16:59Speaker 01

  Codex, Claude, yeah.

• 17:01Greg Isenberg?

  Use one of those to basically say like, okay, for certain tasks, I'm going to be using local models, for certain tasks, I'm going to be using the best-in-class cloud models. And then together, ultimately, you're getting, you know, great results in terms of the output, but you're also not spending through the wazoo. And if you're a token maxi, or like you and me are, like, in the sense of like, we're always pushing into the limit around anything we're building to get the most out of AI, because, you know, we don't want to hire 100 people, 500 people, and stuff like that. It's helpful to do that.

• 17:36Speaker 01

  And exactly, and anecdotally, on two parts, right? One, internally, within our company, you know, I think Satya at Microsoft, you know, mentioned how, like, human capital plus token usage is now a big factor into what they're doing, right? A lot of companies are now moving away from, you know, having direct access to the Claude Code API to run the tokens because of how expensive it's become, right? So they're canceling subscriptions. So we're seeing this firsthand with a lot of companies now are saying, okay, cool, this first year was great, you know, we had the mandate, you know, AI adoption, token maxing, you know, that's how we're going to measure success, and that's how we're going to become AI-native. Now they're like, wait a minute, okay, cool, we've done this, but we're spending way too much money on tokens, how can we now be more effective, right?

• 18:19Speaker 01

  And I'm seeing this firsthand, too, where it's like, especially now, right?

• 18:25Speaker 01

  In a way, you can have some sort of direct ROI between the tokens you're spending within the engineering team, because you're like, okay, cool, we're saving a lot of time, there's an output, you know, engineer's expensive, we get that. But now you're providing the same level of harnesses and models to the non-engineering teams that, you know, are one-shotting, like, hey, help me format this email, and they're using OPUS 4.8 High Thinking. You're like, maybe that's probably not the right model, and that's a governance issue, right? That's a big thing. And I'm having these conversations with companies right now, where they're saying, hey, can you help us figure out, like, how to build governance and proper education on how to actually use the right models? And this is where I think model chaining is a big factor, right? By the way, you know, John at marketing, maybe you shouldn't use OPUS 4.8 to run this like to just format this email for you.

• 19:09Speaker 01

  And just helping them understand that. And I think, you know, I wouldn't be surprised if in a year from now, companies start. And we've been thinking about it as well. We're like, hey, why don't we just get our own machines and start running some local models? Cause it's a lot more effective, especially how much money we're spending on tokens.

• 19:23Greg Isenberg?

  What's the, like, just to play devil's advocate, why wouldn't I just use OpenRouter and call it a day?

• 19:30Speaker 01

  Please do. Yeah. Yeah, absolutely. I think they should.

• 19:33Greg Isenberg?

  What is, because you know, when I'm on, on X, I see a lot of people being like, buy a Mac Studio or buy, you know, these expensive devices. You know, for, for people listening, do they need to buy a local piece of hardware? Like if the price even goes up to X, or should they just use OpenRouter and Cursor? Or, or, or ClaudeCo, you know, whatever harness they, they want.

• 19:58Speaker 01

  If I just have this right machine, I'll get to this, you know, result. That's not how it works. You know, no, you don't need a Mac Mini. You don't need this equipment. You can get started today, you know? And what I love about OpenRouter and all these other tools is, again, they're so agnostic. They make it easy for you to be able to access this in the cloud. They run the models locally and it's credit based. Load $20, get it going and easy to set up. I highly, highly recommend if you're starting to dabble with this and token usage is a thing for you. Get one of these Asian harnesses set up now that there are a lot of them are model agnostic. Run some tokens in OpenRouter, get these open models in there and start, start vibing. Just get some, like, I love to experiment to see how far I can take this.

• 20:39Speaker 01

  What if I plan with OPUS, review with GLM 5.0, execute with 5.2? And then review with Composer 2.5 or Codex 5.5. There's a lot of ways and I think we can be really effective. And I think that's what the smart people are going to be doing in the near future.

• 20:54Greg Isenberg?

  Some people are saying, I don't care how much tokens cost because I think there's so much opportunity in building startups and optimizing and AI arbitrage that I don't even care if it costs me whatever. What do you say to those people who are just basically ignoring this whole open source, local AI movement?

• 21:15Speaker 01

  I used to be the exact same person, you know, in our first episode, you're like, yeah, how much does this all cost? I'm like, I don't know, man, I'm just vibe spending. And I think that mentality has changed now. I can see that my usage limits are being hit faster and my cost is going up. And now that our team is expanding internally as well. So I think as a solo person, it's a lot easier to build a case or rebuttal around why you should just token max as much as possible, which itself is kind of like ironic. You shouldn't be token maxing. You should be token minimizing as much as possible and output maxing instead. So my answer to that is if it works for you and you can directly have an ROI that you can show that, hey, I spent $200 and I got a thousand out, great. Otherwise, sooner or later, the subsidy is gonna run out.

• 22:00Greg Isenberg?

  All right. Well, I think that's, that's the episode, you know, unless there's anything else you want to add before we bounce?

• 22:06Speaker 01

  Yeah, I mean, for people that are trying out, dabble with it, play around, have some, you know, see, see what it can do, at least in the front end for you and start working back in tests. And yeah, I hope they, they got, they learned something from this.

• 22:17Greg Isenberg?

  I'll include links for where to follow Amir. He's always one of my first calls whenever I'm trying out new stuff. And so I'm happy that you were able to jump on. I appreciate you. We appreciate you. Give, give him a follow, like and comment this video. Let us know what you think. We'll be in the, in the comment section. Just, just, you know, out there trying to help and learn and, and thanks a lot, Amir. I'll catch you on the next one.

• 22:43Speaker 01

  Thanks for having me.