>On output tokens, GLM-5.2 is less than a fifth the price of Opus.
Opus is most expensive model in pay as you go model, but IMO fair comparison should include subscription price as well. For example when one has $100 Claude Max and use it up through the month, it might not be more expensive than GLM, or at least not 5x.
I’m currently trying to figure out whether a downgrade from Max 5x to Pro in combination with one of those would save me money and if so, how much.
Edit: seems like Anthropic Pro + GLM Pro (Yearly) would let me almost halve my costs of Anthropic Max 5x. Only concerns are about GLM 5.2 not having vision support and also being kinda slower and also not being as good as Opus.
I'm considering shifting to the OpenAI $20 plan + GLM. OAI has the best computer use, vision support and the best programming intelligence of any model short of Mythos/Fable, and the quota is a lot more generous than the Anthropic $20 plan.
Yes this is true. This test was run on a $20 pro Claude subscription. I would definitely love to try use both models on the highest plans for a whole month and compare the two, great format for a future head-to-head comparison.
Oh I know. But people often point to the API usage cost as an indicator of the magnitude of subsidisation, or to say that the big labs are far less efficient than cheap competitors.
I'm saying that this is not necessarily the case. They do a lot of optimisation and don't have the same price pressure to lower margins. They may not be losing as much on subscriptions as people think.
Oh hm, I've never seen this. API prices have always been exorbitant, I'm sure they're making good margins on that. Let's hope they aren't losing as much on subscriptions, because I'm not ready for everything to be API costs.
Yes. I did not left but visit the site less often and kind of worry about its future. The engagement over what I post is just much lower, while the number of reported visitors seems to increase. I don't mind some good quality AI comment, sometimes one see it, but overall it is slowly becoming ghost town.
I need to do that anyway, but I can treat the agent I work with as a human, I do that for months and didn't encounter any big problem with that approach.
From a moral perspective, I would argue that this is still theft of IP, even if it's a "clean room reimplementation". The code carries valuable information about what works and what doesn't — knowledge that Anthropic had to discover through real work and iteration. It's the same as a Chinese factory duplicating a product: they skipped the entire R&D phase and saved time and money.
But nobody says code is free(?). Certainly not Claude, that experimental compiler costs $20K to build. That openclaw author admitted in Lex Fridman talk that he spends $10k's on tokens each month.
Meta question: do you guys feel the adblockers will maybe not be that important in the future? As for myself, I ended up to use just a few websites, but those are reputable and I don't mind a few ads they provide. The only adblock which is still very much needed is one for Youtube.
I used to run pihole on a Pi and now I directly run unbound, still on a Pi. The difference on a great many sites is night and day: you simply get way fewer ads. And that's just by using a DNS blocklist.
Occasionally I'll get one site that refuses to load because I've got an "adblocker" but most sites do work fine, just with way fewer ads.
I usually now just ask agent, for example Gemini in Antigravity to check certain article or a group of articles, like "check all AI-related article in tldr.tech and tell what is interesting"... I am already a bit lazy to browse myself, and in this process I dont care about ads.
Yes indeed, and also the title promise - I looked forward to read how the personal blogs are back, only to discover the author didnt provide any evidence, but not even examples. Maybe they are indeed back, if we count Substack newsletter archive as a "personal blog".
OpenAI has seemingly done everything they can to put publishers in a position to make this demand, and they've certainly not done anything to make it impossible for them to respond to it. Is there a better, more privacy minded way for NYT to get the data they need? Probably, I'm not smart enough to understand all the things that go into such a decision. But I know I don't view them as the villain for asking, and I also know I don't view OpenAI as some sort of guardian of my or my data's best interests.
But that vault can contain conversation between me and chatgpt, which I willingly did, but with the expectation that only openai has access to it. Why should some lawyer working for NYT have access to it? OpenAI is precisely correct, no matter what other motives could be there.
> We may use Personal Data for the following purposes: [...] To comply with legal obligations and to protect the rights, privacy, safety, or property of our users, OpenAI, or third parties.
OpenAI outright says it will give your conversations to people like lawyers.
If you thought they wouldn't give it out to third parties, you not only have not read OpenAI's privacy policy, you've not read any privacy policy from a big tech company (because all of them are basically maximalist "your privacy is important, we'll share your data only with us and people who we deem worthy of it, which turns out to be everybody.")
> but with the expectation that only openai has access to it
You can argue about "the expectation" of privacy all you want, but this is completely detached from reality. My assumption is that almost no third parties I share information with have magic immunity that prevents the information from being used in a legal action involving them.
Maybe my doctor? Maybe my lawyer? IANAL but I'm not even confident in those. If I text my friend saying their party last night was great and they're in court later and need to prove their whereabouts that night, I understand that my text is going to be used as evidence. That might be a private conversation, but it's not my data when I send it to someone else and give them permission to store it forever.
Listen, man, I willingly did that murder, but with the expectation that no one would know about it, except the victim. Why should some lawyer working for the government have access to it?
Opus is most expensive model in pay as you go model, but IMO fair comparison should include subscription price as well. For example when one has $100 Claude Max and use it up through the month, it might not be more expensive than GLM, or at least not 5x.