I swear, if I read forms with “genuinely” one more time I am gonna scream. FUCK LLM WRITING
i love that word, and now it's genuinely (hehe) ruined. thanks, claude
GLM 5.2 vs. Opus
351–360 of 367 posts
Re: GLM 5.2 vs. Opus
#352Earlier quoted context omitted.
My pi usage over the past ~5 months went roughly like this: * Install pi and a bunch of extensions from their package repo * Realize that all the packages (with a few exceptions) are massively overcomplicated and vibe coded * Ask pi to rebuild a very simple version of the packages I used. So e.g. subagents - all the default subagent extensions are massively complicated with named agents, recursion, communication. I m…
What type of task are you running for ten hours? Is this a programming task? I've not come across a programming task that would take an LLM ten hours.
There's quite a few tasks I've found that work like this, although if course most tasks don't and require a much higher degree of interaction. The prime examples are read only audits of very large codebases, and that's what I was was running overnight. One file per subagent, each subagent writes a report with recommendations. Since it's pi and the subagents have very now scope, looking at them they ranged between 7-40k context use per subagent. I've found codex maxes out at about 50 concurrent subagents before I start getting rate limited, so the coordinating session is instructed to run them in batches of 50. My subagent extension is set up to make this as efficient as possible, the subagents can share a prefix and suffix prompt then a list of name + specific prompt in json format.
Overnight it ran ~800 tiny auditors. I then run synthesis on the written audit files, extract bugs, then another round to find which ones have a common source, group them by priority etc.
I've cautiously started doing larger tasks that are not just read only, for example I was dealing with a large codebase full of lint and type errors, so I sent out waves of workers with clear instructions to only fix obvious/trivial issues to and otherwise to append to a todo file for my review. That worked well and cleared a few thousand issues over several hours.
I don't really want to share any other tasks I've worked on this way because it'll draw out the agentic coding sceptics and I'm not interested in defending my workflow.
Re: GLM 5.2 vs. Opus
#353Earlier quoted context omitted.
My pi usage over the past ~5 months went roughly like this: * Install pi and a bunch of extensions from their package repo * Realize that all the packages (with a few exceptions) are massively overcomplicated and vibe coded * Ask pi to rebuild a very simple version of the packages I used. So e.g. subagents - all the default subagent extensions are massively complicated with named agents, recursion, communication. I m…
I'd like to study your setup. Would you be willing to share? Perhaps a github repo of your 5 extensions or even a pastebin if you would be so inclined. I would be grateful to learn more about this by studying from your success...
I might share it at some point but I think it's quite similar to a lot of others out there, except that it's very specific to my personal projects and goals. If I shared it I'd need to spend at least a while cleaning up and improving docs.
It's one of the reasons I suggest you study the famous setups (oh my pi, or superhuman skills etc.) and convert them to your personal needs.
Re: GLM 5.2 vs. Opus
#354> Through an API it costs a fraction of Opus, and you can run it yourself for free if you have the hardware. I haven't been keeping up on hardware costs for state of the art LLM inference, but this remark made me ask myself how many readers of the article would actually be able to run this model on hardware they own. How much would it cost to acquire such a setup?
GLM-5.2 performing like it would from a good provider - 8x B200s, so $450k. (No personal experience here) GLM-5.2, severely quantised, 512GB Mac Studio, somewhere between $10k-$35k for a used M3. Or run it on a CPU with 768GB of RAM by getting an old PowerEdge with DDR4 for around $5,000. Qwen-3.6-35b-q6, runs well on an RTX 5090 ($4000 + cost of a PC), runs medicore on an Intel Arc B70 ($1000 + cost of a PC plus lot…
Holy AI, this shit is expensive. I was a bit suspect (no experience too) so I run some Claude calculations and it's also giving me a $350-450k to run GLM-5.2 at full precision (un-quantized). For rental on Azure, it's giving me $96–$144/hr. That translates to $22/M tokens which way more expensive than API pricing at z.ai. To get close to API pricing, you have to seek cheaper providers but that only gets you close to z.ai pricing not lower.
Caveat here is that all of this is Claude math, but would be interested in someone more knowledgeable of the math chiming in. I was thinking that API pricing was highly inflated in order to cover subscription costs but with these calculations it might be not?
Re: GLM 5.2 vs. Opus
#355No doubt the open ecosystem is making huge strides, and the gap between open models and the commercial frontier keeps narrowing. What's an Opus 4.8 today will likely be a large open model in a few months — and in a year or two we might have consumer-hardware models matching today's frontier capability. Just look at the recent Qwen and Gemma releases.
It's worth saying the frontier closed labs are charting the path, and the labs releasing open weights are following close behind at a fraction of the cost (though 500B–1T models, open or not, aren't exactly within everyone's reach).
A future where capable AI is genuinely accessible to everyone doesn't look far off — especially since at this point a lot of the robustness and usefulness comes down to the application layer wrapping the model, not just the model itself.
Re: GLM 5.2 vs. Opus
#356Earlier quoted context omitted.
Take a look at Ollama Cloud: https://ollama.com/pricing You get access to a whole bunch of bleeding edge open models including GLM-5.2, Kimi K2.7, DeepSeek 4 Pro, etc. Inference is run on US/SG/EU cloud providers with zero data retention policies. The $20/mo tier is very generous, in my experience.
Well I tried the $20/mo tier and used GLM specifically and did maybe 3-4 hours of work and I'm already through 50% of my monthly tier and blew through my time limited quota twice. I won't renew for another month. Which I think only underscores my point that actually the GLM models are not very cost effective. They essentially cost the same as the SOTA models from OpenAI and Anthropic, while not being quite as smart.…
Ollama can’t meaningfully subsidize their subscriptions - there is no business case to do so because they are a commodity host. If you want to compare subsidized subscription value you would need to compare with z.ai’s plans. One problem with any comparison is that they are all very opaque in terms of usage and the plans change a lot over time. I got on pro at $30 a month so it’s a very good value - compared to $20 Claude/Codex plans I get at least 10x the usage and I use all 3 regularly. At today’s prices Codex pro ($100) is likely a better value.
But if you are building a product or in an enterprise environment where you essentially have to pay API rates then GLM is the best value hands down.
Re: GLM 5.2 vs. Opus
#357Re: GLM 5.2 vs. Opus
#358Re: GLM 5.2 vs. Opus
#359> GLM-5.2 cost a fraction as much. Opus finished in half the time and shipped a cleaner game.
The only thing you measured was a single tunable of how much work should be done on a vague prompt. Now make the prompt be something that causes GLM-5.2 to cost 4x of previous budget, to get something comparable.
(And the wallclock time measures the inference provider, not the model.)
Re: GLM 5.2 vs. Opus
#360Earlier quoted context omitted.
The docs are really helpful : https://dev.synthetic.new/docs/guides/claude-code Glad you figured it out :) Let me know your thoughts about the quota and GLM5.2, so far I don't think I've came across against anything better, $/usefulness wise.
This whole thread sounds like an advert.
lol, I ended up not signing up, got lazy and yeah I agree, there's been a lot of that on HN these days which makes things feel weird