Rendered at 08:54:07 GMT+0000 (Coordinated Universal Time) with Cloudflare Workers.
gonzalohm 21 hours ago [-]
Why tech companies never price things according to what they really cost. This new strategy of entering the market at a loss and then cranking up prices is getting old.
Anon1096 20 hours ago [-]
Nothing is priced according to what they really cost. Things are priced at what the market will bear based on supply and demand. You might as well ask this question for oil, iphones, bitcoin, restaurants, or housing the answer is all the same.
gonzalohm 20 hours ago [-]
I have been paying the same for my Kagi subscription since the start
shlewis 4 hours ago [-]
Which is almost certainly less than how they really should be, according to what they (allegedly) cost.
Various cost-saving measures have been discussed, including a separate Assistant plan.
Again, it will be priced according to how much the market will deem fit, which is less than what it costs at the moment.
dumberquestions 11 hours ago [-]
Predatory pricing is a separate issue from supply and demand.
partsch 21 hours ago [-]
To attract users :)
--
The question is probably more: How can it be legal to offer a service without specifying exactly what you're getting?
Levitating 20 hours ago [-]
You can still find honest companies providing correctly priced services. Purelymail.com handles my email. There are many VPS hostars, cloud storage providers etc with honest pricing.
KellyCriterion 21 hours ago [-]
This is not limited to tech companies!
Underpricing is a very often applied method to enter/grow/develop a market.
cma 21 hours ago [-]
When a new restaurant opens in a college town, there will often be things like an opening day where everyone with a student card gets a free taco or whatever. Cheaper than buying a physical ad at the football stadium, and targets the people that might end up eating there multiple times a week (developers bringing Claude to their companies) rather than a bunch of people just in town for the game (Claude TV ads).
The training data flywheel though was probably as or even more important at least as of whenever it was that they switched it to opt out instead of opt in on subsidized subscriptions.
gonzalohm 20 hours ago [-]
Right, and while you are getting your free taco you can probably ask for a copy of the menu so you can know if you are able to afford it
One thing is offering free demos or discounted products for a short time and a different one is making your consumers believe that the current price is the long term price.
I just want transparency. Are they done increasing prices?
cma 13 hours ago [-]
> I just want transparency. Are they done increasing prices?
The initial announcement said it would only be available to subscriptions for a few weeks. The extra usage they said was for a limited time, I think it was 50% extra and they now are reducing that by 17%, less reduction than they had said.
Because it's a monthly plan they usually have given heads up of at least a month on stuff.
The API prices outside of personal subscription have always been more. Personal subscriptions are likely there for data scraping of company codebases (when users don't opt out), and for spreading the product to then get picked up by businesses the users work at or other word of mouth for API prices which they have said are profitable.
sieve 15 hours ago [-]
I subscribed to Claude Pro for one month earlier this year. The subscription lasts as long as you chat with it in the UI. The moment you use Claude Code, it becomes useless and you keep hitting limits. So, great for hammering out a specification, not so great for actually writing and debugging code.
I have a lot of hobby projects and cannot afford to drop $200/m for assistance from LLMs with above average intelligence. The Chinese models work quite well, and are cost effective, in comparison. My usage over the last few months:
- Gemini / AI Studio = translations
- Kimi 2.5 / OpenCode Go = Compiler/VM project, AI for some board games, a new document format and compiler for it,
- DS4 Flash / DeepSeek (had some credits lying around) + OpenCode Go = Template-based OCR engine
- MiMo 2.5 and MuseSpark 1.2 Contributor = The OCR engine + a product I am brainstorming + a lot of single page tools
If you select models judiciously, and can write good AGENTS.md and extremely short but potent ARCHITECTURE.md documents, the $10 you pay to OpenCode Go will last a long while even when you consume a hundreds of millions of tokens (mostly cache hits) a day. You can get multiple models to perform code review of modules against the ARCHITECTURE you have specified and get other models to execute the fixes.
nottorp 12 hours ago [-]
> and can write good AGENTS.md and extremely short but potent ARCHITECTURE.md
It does for me. You need to ensure that they are injected into context. Your models might be broken if they ignore it. As good as ignoring a prompt.
nottorp 3 hours ago [-]
> Overall, our results suggest that context files don’t improve coding agent performance, and should only contain specific additional instructions beyond what is already available in the codebase
That's what the study says.
I recently did a much smaller scale test. Some coworker was pestering me about some brilliant skill file so i backed up Claude's state and told it to make a plan for some refactoring task with and without those skills installed.
The results were identical. I didn't measure token use though.
sieve 51 minutes ago [-]
[dead]
xlayn 19 hours ago [-]
I pay a $100 subscription which feels infinite for me, even using only fable... 2 weeks ago I started getting notifications of running out of credit... to me it feels more like 30% than 17%...
OHHHH and surprise, I'm in my 50% boosted/higher limits... so this means I'll probably hit the $100 limit now constantly... I use the thing a bit... yesterday and it shows 14% of the week allocation used...
zcw100 16 hours ago [-]
I worried about this myself. 50% higher limits were nice but the flip side is you get used to it and then it just feels like they're taking 50% away. I'm not sure it's going to have the effect that Anthropic was expecting....or maybe that was the point. Who knows. I can see both perspectives.
BugsJustFindMe 19 hours ago [-]
Meanwhile OpenAI resets everyone's weekly Codex limits back to 100% seemingly every other day as a kind of game.
vablings 15 hours ago [-]
I am very tempted to jump over to Codex for a little while. My friends are split between codex and claude code and hearing them getting a new reset every day for the past week has been giving me FOMO. Maybe that is intentional
ByteOfWood 16 hours ago [-]
Isn't this effectively the same thing Anthropic was doing? Eventually the resets will stop coming too, either entirely or just less frequently.
alasano 16 hours ago [-]
Yesterday was a funny one, they announced in advance that a reset would happen at 6PM PST and encouraged everyone to burn through as much usage as they could.
JangoCG 23 hours ago [-]
Glad that I canceled my sub
freepiai 20 hours ago [-]
[flagged]
throwa356262 23 hours ago [-]
What is going on? Why are everyone except OpenAI short on compute?
Even Chinese ultra cheap providers are slowly reducing token plans or increasing prices.
cedws 23 hours ago [-]
I doubt it’s compute, Anthropic just bought a huge amount of capacity from xAI. I mean SpaceX. No, I mean SpaceXAI.
They’re trying to stop bleeding cash.
tcp_handshaker 21 hours ago [-]
That data center was setup wrongly, is broken and unusable for training. Its amazing Musk manages to fool even Anthropic.
skeledrew 18 hours ago [-]
Amazing that Anthropic is using this "broken and unusable" data center successfully, pushing new models and providing sufficient inference that we no longer have the absurd limits from the weeks leading up to them making that deal.
dgellow 22 hours ago [-]
Anthropic will IPO soon and need to show good numbers. Everything happening in the next weeks/months is about that.
The others are trying to reduce their costs, you have to understand that nobody has a moat, and nobody is profitable. While at the same time the cost of energy and hardware is going up.
gmerc 23 hours ago [-]
Everyone has been running promotions and, gosh, demand is up.
csomar 22 hours ago [-]
It’s not clear how much each token costs and how much it’s being subsidized. It was never meant to last, at some point the music was going to stop and you have to pay full price.
anthonypasq 17 hours ago [-]
api pricing has 80+% margin.
skeledrew 18 hours ago [-]
I have no problem here. If I notice that I'm getting markedly less than before, I'll just switch to Z.ai/GLM. Been on the edge for a while now just waiting for something to push me over.
activeing123 18 hours ago [-]
[flagged]
bschwarz 1 days ago [-]
Interesting wording:
[...] that will make it feel like you’re getting more from Claude
pllbnk 1 days ago [-]
Well, to be fair, you cut the 6 important words that show the context:
> We’re working on exciting changes that will make it feel like you’re getting more from Claude
On the other hand, I think the biggest improvement they could do is allow using Claude Code subscriptions from third party harnesses with no strings attached.
xerios 24 hours ago [-]
There's a difference between making you feel and actually improving something. The word choice is definitely suspicious.
pllbnk 22 hours ago [-]
AI industry does everything on vibes :)
xingped 22 hours ago [-]
I think the problem is with the word feel. Reducing how much you get while saying it'll feel like you're getting more isn't usually a good thing. It's very weasely wording hoping most people will miss that and think what they said was that they're getting more, not that they'll _feel_ like they're getting more.
wtfwhateven 18 hours ago [-]
What strings are attached? I thought they reverted the API usage changes
Where did you find the information about 3rd party harnesses?
la_oveja 1 days ago [-]
hes saying he wishes they would do it
pshirshov 19 hours ago [-]
Ah, my bad.
nusl 24 hours ago [-]
Yeah, because it's 17% when compared to the current +50% "promotion" they have going. They're raising base weekly limit by 25%. It's still more than non-promotion usage limits.
pjerem 23 hours ago [-]
Still bullshitting numbers so much that one does not really know what it pays hundreds/month for.
yonatan8070 19 hours ago [-]
Also:
> we're permanently raising standard weekly limits in Claude Code by 25%
lostNFound 10 hours ago [-]
At risk of getting defenestrated (or worse, expelled...) getting a permanent 25% increase sounds good over a 50% temporal increase. Imagine if this was your wage you were talking about, wouldn't you rather the 25% perm?
I use only Fable High as well. 8 hour workday, it does my job, straight up. I only supervise. Mostly frontend and in particular, mobile, where Fable (Claude in general) is quite better than the competition, in my opinion of course.
I also use it for managing a company's website separate from my job, and my own website. And for daily studying of an unrelated university career + daily personal use. For all the ones in the current paragraph I use Opus 5 xhigh.
I don't think I've ever gotten even to 50% of the (currently augmented) levels. A 25% permanent increase is good in my books.
I mean, what? I just got the memo that we're in operation Complain and Destroy. This is horrible! I will now go buy my Torch&Pitchfork two-pack for 17% off.
joshstrange 21 hours ago [-]
Another week, another case of getting jerked around by Anthropic… I’m tired of my $200 subscription changing or threatening to change every 2 weeks or less. Not that I think OpenAI and “resets” is a life I want to live either. I guess we will see if they go through with this or if the backlash stops them, again.
Various cost-saving measures have been discussed, including a separate Assistant plan.
Again, it will be priced according to how much the market will deem fit, which is less than what it costs at the moment.
--
The question is probably more: How can it be legal to offer a service without specifying exactly what you're getting?
The training data flywheel though was probably as or even more important at least as of whenever it was that they switched it to opt out instead of opt in on subsidized subscriptions.
One thing is offering free demos or discounted products for a short time and a different one is making your consumers believe that the current price is the long term price.
I just want transparency. Are they done increasing prices?
The initial announcement said it would only be available to subscriptions for a few weeks. The extra usage they said was for a limited time, I think it was 50% extra and they now are reducing that by 17%, less reduction than they had said.
Because it's a monthly plan they usually have given heads up of at least a month on stuff.
The API prices outside of personal subscription have always been more. Personal subscriptions are likely there for data scraping of company codebases (when users don't opt out), and for spreading the product to then get picked up by businesses the users work at or other word of mouth for API prices which they have said are profitable.
I have a lot of hobby projects and cannot afford to drop $200/m for assistance from LLMs with above average intelligence. The Chinese models work quite well, and are cost effective, in comparison. My usage over the last few months:
- Gemini / AI Studio = translations
- Kimi 2.5 / OpenCode Go = Compiler/VM project, AI for some board games, a new document format and compiler for it,
- DS4 Flash / DeepSeek (had some credits lying around) + OpenCode Go = Template-based OCR engine
- MiMo 2.5 and MuseSpark 1.2 Contributor = The OCR engine + a product I am brainstorming + a lot of single page tools
If you select models judiciously, and can write good AGENTS.md and extremely short but potent ARCHITECTURE.md documents, the $10 you pay to OpenCode Go will last a long while even when you consume a hundreds of millions of tokens (mostly cache hits) a day. You can get multiple models to perform code review of modules against the ARCHITECTURE you have specified and get other models to execute the fixes.
https://pivot-to-ai.com/2026/08/27/your-agents-md-file-doesn...
That's what the study says.
I recently did a much smaller scale test. Some coworker was pestering me about some brilliant skill file so i backed up Claude's state and told it to make a plan for some refactoring task with and without those skills installed.
The results were identical. I didn't measure token use though.
Even Chinese ultra cheap providers are slowly reducing token plans or increasing prices.
They’re trying to stop bleeding cash.
The others are trying to reduce their costs, you have to understand that nobody has a moat, and nobody is profitable. While at the same time the cost of energy and hardware is going up.
> We’re working on exciting changes that will make it feel like you’re getting more from Claude
On the other hand, I think the biggest improvement they could do is allow using Claude Code subscriptions from third party harnesses with no strings attached.
> we're permanently raising standard weekly limits in Claude Code by 25%
I use only Fable High as well. 8 hour workday, it does my job, straight up. I only supervise. Mostly frontend and in particular, mobile, where Fable (Claude in general) is quite better than the competition, in my opinion of course.
I also use it for managing a company's website separate from my job, and my own website. And for daily studying of an unrelated university career + daily personal use. For all the ones in the current paragraph I use Opus 5 xhigh.
I don't think I've ever gotten even to 50% of the (currently augmented) levels. A 25% permanent increase is good in my books.
I mean, what? I just got the memo that we're in operation Complain and Destroy. This is horrible! I will now go buy my Torch&Pitchfork two-pack for 17% off.
https://news.ycombinator.com/item?id=49491282