I can't help wondering what kind of models we'll see coming out of China once it gets its own chip fabs up and running. Right now it sounds like the US's export ban is not slowing them down a whole lot.
With subsidization from the Chinese government they will probably be equal to or better than the models here. I mean, have you looked at the author list of any given AI paper published within, say, the past 5 years? I wouldn't be surprised if half or more AI researches are from China.
There does not seem to be a big penalty for going slow anyways. People seem to just switch on cost as soon as a model can do a task well enough. There do not seem to be strong network effects or vendor lock in.
Seems to me that going slow is the better long term tactic. China can just let the USA pay the high R&D costs to figure out what works, then just copy what works.
Open weight models from Chinese labs tend to be significantly cheaper.
I think theyre absolutely needed. I can't afford 200 USD a month for personal use of coding AI, and I don't think such prices are reasonable for most of the world economy anyway. Not to mention US firms might be giving their employees a lot more than that.
It's increasingly feeling, to me, that theres a gap building up between haves and have nots. But then, we get news of these open weight models that are reasonably priced in inference with reasonable capabilities. Yes, they take maybe 6-9 months to get there, tbh, that's not a bad trade off at all.
DeepSeek through their own API has saved me tons of tokens honestly. Even though it is not as smart as Kimi or Claude, their level of entry is very low with a top up of 2$ and Pay as you go compared to the subscription of Claude or 20$ top up of Kimi
For personal use I’m considering using the frontier models from openai or anthropic to create a plan with research and brainstorming etc with enough details for cheap models to be able to follow (glm, deepseek etc) - with openrouter - will monitor how cheap and effective that turns out to be.
If we can agree that the AI model is at least as capable as a junior engineer or new contractor, how’s that different to saying “software engineering isn’t worth $200 a month”?
Has a very race-to-the-bottom feel to it.
Though in the grand scheme of it, $200/mo probably isn’t the real price either. Also looking at it not just in a vacuum - paying for a product that can change what you get from under you doesn’t seem great anyway.
At least with a locally-hosted model you know what you’re getting.
The appropriate price is what the output is worth to you. Some people could pay $10,000/month, some $5 and feel like they were breaking even. There is a big jump between convenience and curiosity uses versus business critical.
OpenAI already charges enterprise users a premium purely for that title over on-demand, no-contract usage. Retail users get a good deal. People make a lot of hay about subsidies but this is a very sane approach if you want exposure to these three different types of customers.
Significantly cheaper than comparable models if you are using openrouter [0]. Just yesterday I spent roughly 13 cents centering some divs using Deepseek in a personal project. It would have been north of $1 to do that with a US frontier model.
Not that it matters but most of the open weight models aren’t actually censored that way: they run another layer on top of to do that. At least some of them do, Step 3.7 Flash locally happily tells me about the Tiananmen Square massacre
You made me realize something. I routinely spend upwards of 500$ per month on LLMs for coding (expensed towards clients). However I live in a place where 500$ is around the avg. salary. I’m lucky that I know my way around western clients. Clients who pay these expenses and are happy to work with me because I am still about 50% cheaper than local talent in EU/US, while my salary at home converts to an upper class income at the highest tax bracket.
Which of course causes some unfairness on both ends. Nobody here can compete with me. I often use left over tokens on local client projects; which despite lower pay, still pays off because they now take hours not days or weeks to complete. And nobody in the local clients talent pool can compete with me; unless they charge about half the market rate.
Take away my 500$ monthly grant; and I’d be more or less screwed. Better open models will more or less start to reduce this advantage. It’s not like I positioned myself here on purpose. But it’s definitely a „right place, right time“ situation.
AI is the first technology that doesn't incentivize offshoring, and incentivizes co-location of talent.
A NYC dev and a dev in india have the same ai costs, based the ratio tokens/salary it becomes less of comparative disadvantage to be in NYC.
Now combine that with the fact that AI makes the act of generating code less a % time of the job, and the ability to get/refine requirements more of the job and you have a decent shift.
The tokens/salary ratio is not relevant at all. Because while 200-500$ is a lot of money, it’s still a fraction of the salary you’d pay any dev in the world. It just comes out as a tooling expense. It also matters how those devs use the tools; you can’t assume everyone gets the same out of it. So that amount can last a day or it can last a month. I would say a dev in a developing nation would be more budget aware than someone being used to everything being priced in NYC rates.
For example I build other AI products and I have been hyper aware of the token spend of our users. I was going crazy seeing that some users were having 5$ conversations. So that was optimized and I found ways to use sub agents to get it down to 1-2$. Just for management asking me why I was worrying to begin with? The users using these are consultants being paid 120$ per hour. They have a daily 10-20$ token expense, no problem. “But amazing job on the cost reduction.”.. well 5$ for me is what I spend on food daily. While the consultant is slamming: “yes” 10 times in a chat , for whatever reason for the same cost. Would the NYC dev care as much natively? No.
You can still hire three devs in India for the price of a dev in NYC. Now you give them AI and you might only need 1-2. That makes offshoring even more appealing, not less. And the dev in India now having tooling to out compete local talent. Well that’s my reality (I am not in India though).
The problem is that the differences between flagship and local models are compounding heavily. An 4% different could be massive when you keep iterating on the same code base.
I read these stories and I can never figure out how people are managing to use these $200 plans. If I really go full bore, I can sometimes max out the $20 plan. Even then, it already produces more code than I can reasonably review and merge.
do you do it for a job (8 hours a day)? and do you work in large, mature projects (more than 5 team members)? A big part of it is dealing with frankly terrible architecture and 15 people's different ideas of how things should work (and the spam theyve been able to do with their own agents makes this worse)
With open weight models there is true inference competition. Whoever can serve the model at the lowest price. And the consumer wins. Capitalism, served by China.
The tokens cost the same everywhere on earth. This does hurt some cost advantages of outsourcing when tokens start to become a bigger part of development costs.
As much as I don't like Mark Zuckerberg, part of me wishes he would get his head in the game and compete with these models, he's literally got all the capability to do so, and he could easily sell the model through deals with GCP, AWS, and Azure. Hell, Amazon needs a hot model they can host that's exclusive to them I feel like, maybe he can work something out with them, whatever the case, it seems so glaringly obvious to me, I'm not sure why he hasn't taken a stab at competing with Claude Code or at least frontier open models and then cutting a deal with cloud providers to recoup the costs of maintaining said models.
He's sitting on a frontier model letting it burn a hole in his wallet that could actually pay for itself.
Kind of funny that you're assuming that you are not paying with your data in both cases.
Do I need to remind you how LLMs are being trained? ...or that Anthropic claimed their codebase is 100% vibecoded, making it uncopyrightable by their own logic? ...or that Anthropic took down all Claude Code leaks they could find using DMCA takedown notices? ...or how do you think the caching mechanisms work when there's allegedly no data stored to be able to cache it?
I'm just saying. Anything you build with online models is their training data anyways. Assuming otherwise is pretty stupid at this point.
I just tested GLM 5.2 out via Z.ai in pi for a little one-off project that was already scoped. It actually did a relatively decent job starting out, and figured important things out from context.
But the reasoning traces became increasingly hilarious, with it getting confused and going in loops, doubting itself. I began to feel almost sad, it was like listening to the internal monologue of someone with anxiety disorder.
It made pretty good progress but wound up going in a lot of goofy loops and doing things a bit "off" from standards I'd hoped it would infer, and finally started going a bit nuts, "This is very confusing.", "OH WAIT", seemingly hallucinating a whole side-quest that didn't make sense and looking at making internal system changes to try to achieve its (now very confused) goal when I pulled the plug.
Without seeing the reasoning traces from Claude/GPT it's hard to really know, but it definitely didn't feel like the same quality of reasoning, even if dogged persistence does wind up actually working eventually.
I have a hilarious theory why GLM (and Kimi) have this thinkslop,
apparently Chinese language as token is more information dense than English, so having these wasteful thinkslop in Mandarin isnt that damaging. So the developer focus mostly in Mandarin and didnt think of handling these thinkslop while American AI labs do.
Ive been using glm5 since its release and still prefer it to glm5.1 and so far to glm5.2
Perhaps it is just my harness and workflow, but the older model still seems to work better. Also the token cost is significantly lower. I rarely spend more than $20 a week with $50 cap. Not even half claudes ambiguous minimum $200 a month plan.
Can people share their GLM and open model setups in general please? What provider do you use. Why do you trust it with serving full quality? What harness do you use? Why do you trust it not to have malware (most harnessed are TS apps). I am just trying GLM 5.1 from Nvidia build in open code would love to hear how you all do it, thanks.
I've been using GLM 5.2 recently (company hosted, for non-coding tasks) and it's been strong and reliable. There are areas where GPT 5.5 and Opus 4.x still feel marginally better but only marginally. For most tasks if GLM 5.2 is the only model I have to use I'm productive and happy. This was not true before GLM 5.2. No doubt in my mind that the gap is closing quickly and for most tasks that are not very specialized open models will be usably on par on flagship closed models and have an edge factoring in cost.
For coding I still use 5.5 w/ Codex and prefer that to other models + harness combinations.
It feels like the gap is closing from an intelligence perspective. Or at least doing some kind of log flattening.
Been playing with GLM 5.2 in different contexts. It's less good if you don't max out thinking, but as xhigh it's been able to solve most problems I was throwing at Opus in the about the same amount of time (via OpenRouter).
I've been working with Deepseek V4 Flash (with opencode as the harness). It's been almost indistinguishable from Codex / Claude Code for me. I'm sure I'll run into problems when I get to a stickier ticket to tackle. But so far, it's been quite good, and I find it writes straightforward code.
I do think the Chinese models are good enough for an 80/20 rule use case.
I also use DeepSeek v4 flash and v4 pro, but I can’t settle between using Claude Code or OpenCode and it seems like I waste time switching back and forth (especially keeping my personal SKILLs files synced). On one hand, a ton of engineering work has gone into Claude Code, on the other hand all Chinese models I have tried with OpenCode seem well configured out of the box.
I was thrilled to have Gemini Ultra for a month and use as many Opus tokens with AntiGravity as I could use, but I am happier using less capable models like DeepSeek knowing that it is more fun to do more of the work myself, it is a smaller hit on the environment, and incredibly cheaper.
We have switched approx 80% of our work to deepseek, and it works great. Our setup is a bit unconventional though, we upload all cot / sessions to shared storage and generate centralised project level context. We've found this is helpful in directing and working with these slightly less sota models and getting great value for ai spend.
I'm planning to open source all this infra soon, hopefully useful for others too.
1. SWE-bench Pro
Model Score (%)
GLM-5.2 62.1
GLM-5.1 58.4
Claude Opus 4.8 69.2
GPT-5.5 58.6
Gemini 3.1 Pro 54.2
2. Terminal-Bench 2.1
Model Score (%)
GLM-5.2 81.0
GLM-5.1 63.5
Claude Opus 4.8 85.0
GPT-5.5 84.0
Gemini 3.1 Pro 74.0
3. NL2Repo
Model Score (%)
GLM-5.2 48.9
GLM-5.1 42.7
Claude Opus 4.8 69.7
GPT-5.5 50.7
Gemini 3.1 Pro 33.4
4. DeepSWE
Model Score (%)
GLM-5.2 46.2
GLM-5.1 18.0
Claude Opus 4.8 58.0
GPT-5.5 70.0
Gemini 3.1 Pro 10.0
5. ProgramBench
Model Score (%)
GLM-5.2 63.7
GLM-5.1 50.9
Claude Opus 4.8 71.9
GPT-5.5 70.8
Gemini 3.1 Pro 39.5
6. MCP-Atlas
Model Score (%)
GLM-5.2 77.0
GLM-5.1 71.8
Claude Opus 4.8 77.8
GPT-5.5 75.3
Gemini 3.1 Pro 69.2
7. Tool-Decathlon
Model Score (%)
GLM-5.2 48.2
GLM-5.1 40.7
Claude Opus 4.8 59.9
GPT-5.5 55.6
Gemini 3.1 Pro 48.8
8. Humanity's Last Exam
Model Base Score (%) Score w/ Tools (%)
GLM-5.2 40.5 54.7
GLM-5.1 31.0 52.3
Claude Opus 4.8 49.8 57.9
GPT-5.5 41.4 52.2
Gemini 3.1 Pro 45.0 51.4
Seems to be handily beating Gemini 3.1 Pro. What _is_ Google DeepMind doing (other than bleeding talent to A\ ) ?
It's by far the most competent open model I've tried yet. It's a bit slower than Claude, but in terms of coding capability it seems to get comparable results at least for the work I'm doing.
While I agree with the post in its entirety, I think it would have been worth mentioning DeepSeek V4 Flash as well, which, in my view, had already reached a sufficient, if not high-level of agentic coding before GLM 5.2 (see DwarfStar).
I know very little about the current state of replacability of Opus but I do sometimes imagine a reality where Opus has been rebuilt as an open model. What plan does Anthropic have when it does happen?
Will they still rent out their own model, will they support the open model and become a resource provider? Will they be able to repay the billions of dollars ?
This is probably the first question I would ask someone from Anthropic, if I ever meet one.
75 comments
[ 2061 ms ] story [ 2668 ms ] threadIt may wind up being a massive boost to them in the long run, even.
Necessity is the mother of invention.
Seems to me that going slow is the better long term tactic. China can just let the USA pay the high R&D costs to figure out what works, then just copy what works.
Just costing them a lot more money as they pay multiples more buying on the underground grey market.
I think theyre absolutely needed. I can't afford 200 USD a month for personal use of coding AI, and I don't think such prices are reasonable for most of the world economy anyway. Not to mention US firms might be giving their employees a lot more than that.
It's increasingly feeling, to me, that theres a gap building up between haves and have nots. But then, we get news of these open weight models that are reasonably priced in inference with reasonable capabilities. Yes, they take maybe 6-9 months to get there, tbh, that's not a bad trade off at all.
Has a very race-to-the-bottom feel to it.
Though in the grand scheme of it, $200/mo probably isn’t the real price either. Also looking at it not just in a vacuum - paying for a product that can change what you get from under you doesn’t seem great anyway.
At least with a locally-hosted model you know what you’re getting.
OpenAI already charges enterprise users a premium purely for that title over on-demand, no-contract usage. Retail users get a good deal. People make a lot of hay about subsidies but this is a very sane approach if you want exposure to these three different types of customers.
0. https://openrouter.ai/compare/z-ai/glm-5.2/anthropic/claude-...
If that was true, they would be collaborating with each other and opening up all the results from their work.
People speak of a permanent underclass.
https://www.nytimes.com/2026/04/30/opinion/ai-labor-work-for...
Which of course causes some unfairness on both ends. Nobody here can compete with me. I often use left over tokens on local client projects; which despite lower pay, still pays off because they now take hours not days or weeks to complete. And nobody in the local clients talent pool can compete with me; unless they charge about half the market rate.
Take away my 500$ monthly grant; and I’d be more or less screwed. Better open models will more or less start to reduce this advantage. It’s not like I positioned myself here on purpose. But it’s definitely a „right place, right time“ situation.
A NYC dev and a dev in india have the same ai costs, based the ratio tokens/salary it becomes less of comparative disadvantage to be in NYC.
Now combine that with the fact that AI makes the act of generating code less a % time of the job, and the ability to get/refine requirements more of the job and you have a decent shift.
For example I build other AI products and I have been hyper aware of the token spend of our users. I was going crazy seeing that some users were having 5$ conversations. So that was optimized and I found ways to use sub agents to get it down to 1-2$. Just for management asking me why I was worrying to begin with? The users using these are consultants being paid 120$ per hour. They have a daily 10-20$ token expense, no problem. “But amazing job on the cost reduction.”.. well 5$ for me is what I spend on food daily. While the consultant is slamming: “yes” 10 times in a chat , for whatever reason for the same cost. Would the NYC dev care as much natively? No.
You can still hire three devs in India for the price of a dev in NYC. Now you give them AI and you might only need 1-2. That makes offshoring even more appealing, not less. And the dev in India now having tooling to out compete local talent. Well that’s my reality (I am not in India though).
Also if you run the “loops” they’re now yapping about, it will burn through enormous amounts of usage as well.
He's sitting on a frontier model letting it burn a hole in his wallet that could actually pay for itself.
Do I need to remind you how LLMs are being trained? ...or that Anthropic claimed their codebase is 100% vibecoded, making it uncopyrightable by their own logic? ...or that Anthropic took down all Claude Code leaks they could find using DMCA takedown notices? ...or how do you think the caching mechanisms work when there's allegedly no data stored to be able to cache it?
I'm just saying. Anything you build with online models is their training data anyways. Assuming otherwise is pretty stupid at this point.
But the reasoning traces became increasingly hilarious, with it getting confused and going in loops, doubting itself. I began to feel almost sad, it was like listening to the internal monologue of someone with anxiety disorder.
It made pretty good progress but wound up going in a lot of goofy loops and doing things a bit "off" from standards I'd hoped it would infer, and finally started going a bit nuts, "This is very confusing.", "OH WAIT", seemingly hallucinating a whole side-quest that didn't make sense and looking at making internal system changes to try to achieve its (now very confused) goal when I pulled the plug.
Without seeing the reasoning traces from Claude/GPT it's hard to really know, but it definitely didn't feel like the same quality of reasoning, even if dogged persistence does wind up actually working eventually.
apparently Chinese language as token is more information dense than English, so having these wasteful thinkslop in Mandarin isnt that damaging. So the developer focus mostly in Mandarin and didnt think of handling these thinkslop while American AI labs do.
Perhaps it is just my harness and workflow, but the older model still seems to work better. Also the token cost is significantly lower. I rarely spend more than $20 a week with $50 cap. Not even half claudes ambiguous minimum $200 a month plan.
OpenCode works fine, i just find it very resource intensive for no good reason.
For coding I still use 5.5 w/ Codex and prefer that to other models + harness combinations.
Been playing with GLM 5.2 in different contexts. It's less good if you don't max out thinking, but as xhigh it's been able to solve most problems I was throwing at Opus in the about the same amount of time (via OpenRouter).
Wild time to be alive.
I do think the Chinese models are good enough for an 80/20 rule use case.
I was thrilled to have Gemini Ultra for a month and use as many Opus tokens with AntiGravity as I could use, but I am happier using less capable models like DeepSeek knowing that it is more fun to do more of the work myself, it is a smaller hit on the environment, and incredibly cheaper.
I'm planning to open source all this infra soon, hopefully useful for others too.
I subscribed to their max plan to try it out. It counted me 700M tokens and drained my weekly quota in under 2 days.
Quota just reset less than 24h ago and i'm already >60% weekly quota usage.
For reference the kind of work i did would have used somewhere between 3% and 5% of Codex max or Claude max.
The model is good, the plan is a scam
"Make a pac-man game in a single html page"
It went off and argued with itself for 20 minutes about how to lay out the map and then timed out.
Is 2 better than x.ai
Will they still rent out their own model, will they support the open model and become a resource provider? Will they be able to repay the billions of dollars ?
This is probably the first question I would ask someone from Anthropic, if I ever meet one.
Anthropic rents GPUs from xAI to run Claude. If there's an open weights competitor to Opus, why wouldn't Elon host it directly?