notatoad

if you've got an enterprise contract, doesn't that include pricing? a temporary discount on the base API rate probably isn't super relevant to that.

if you're picking AI models for long-term sustainability you're doing it wrong. There's really no point in locking in model choice for anything more than a month or two these days.

what the spec says doesn't really matter though, the spec is just nonsense people wrote down to try to back-explain decisions that the browsers had already made.

pixels are pixels, and no browser implements them as a fractional-inch measurement. in every actual implementation i'm aware of, they're a hardware based unit.

px is still a useful unit. the user's operating system applying a scaling fator doesn't make it not useful.

if the user's operating system is configured to draw pixels at a 1:1 ratio, your pixels will be drawn at a 1:1 ratio. if the user's operating system is configured to draw pixels at a 2:1 or 3:1 ratio, you as a web developer don't get to override that, but you're still addressing pixels directly. if you want to write software that manages the user's hardware directly without giving them options to override it, the web is probably not the right platform for you.

>reaches out to service.com/agents.md for a fresh copy every so often and uses that to resolve the relevant tool calls.

that is basically what MCP is. except it answers all the questions that your version handwaves away - how often do you get a fresh copy, how do you describe the relevant tool calls, how are the tools organized, and how does auth work.

it's not easier for agents to work with. it's easier for organizations to work with.

for agents, they're essentially the same thing - remote endpoints, and instructions on how to call those endpoints. what MCP brings is centralized updating and distribution of the instructions, and a promise that the skill and the REST api won't be out of sync with each other.

the one thing that skill.md+REST doesn't solve is how you get that skill.md to somebody else's computer, and how you ship an update to somebody else's computer once they've got a copy of the skill. if that's a problem you need to solve, you can either start inventing skill.md distribution protocols, or you can just use MCP.

Lots of smart people have tried to implement micropayments like this, and it has never worked.

Publishers would rather have the chance at $10 than give up the chance at $10 for 10c. And readers don’t want to be thinking about their credit balance every time they read an article.

yes, but the GLM subscription gives you barely more usage than you'd get just paying API rates. it's not subsidized like the claude or chatgpt subscriptions are.

i don't think that's an issue. it's still a perfectly good rule.

use what you're familiar with, until it stops working. then use something else. postgres just goes a lot further than a lot of other tools before you get to the "use something else" phase. and postgres is the database a lot of people are familiar with.

if your highest performing ad is against searches for literally the title of your book, isn't that a signal that you don't need to buy ads? people are searching for your book. they will buy it whether you advertise or not. so why by the ad?

this is framed as amazon doing something wrong, but it sounds like what's actually happening is that his marketer is wasting money. of course amazon is going to accept money if you give it to them for the things they're already doing for free. why shouldn't they?

no, at subscription prices claude is a better value than GLM.

They're only a better value if you're paying API rates

I think I generally agree with you - OpenAI and anthropic will probably succeed here. If the ai bubble pops, they’ll come out on top.

I just think the whole “moat” discourse is silly. OpenAI and Anthropic’s success depends on the same thing every business’s success depends on: their customer base, and their continued delivery of services their customers want to pay for. Not their tech or their compute or anything else. They have no moat because moats aren’t a thing.

Isn’t almost all of anthropic and OpenAI’s compute leased?

If compute is the moat, that doesn’t really make their position any less precarious.

probably not all that much... the market would dip, just like every time a new open weights model gets announced. but hundreds of millions of people aren't going to immediately self-hosting their own models.

the biggest winner in that scenario would be ai providers, who suddenly have a capable model that they can serve much more efficiently. and the incumbents have a whole lot of compute. wouldn't anthropic and openAI just start offering that open weights model at prices that nobody else could compete with?

$100b is payment volume, not revenue. Stripe gross revenue from those payments would be on the order of tenths of a percent of that.

no. i don't know if openrouter even guarantees consistent protocl across models.

the value of openrouter is it offers centralized billing. you can route your calls to any provider you want, test a whole bunch of models against each other, and you just get one bill from openrouter. switching to a new model, or a new provider of the same model, doesn't mean setting up a new billing account with a new provider.

this all seems to fall into trap of most UI design writing, where people write blog posts about what users want, what users think, and what users understand, without actually doing any user testing.

unlike everybody else in this comment section, i like the two-mode toggle. but that's just my opinion. without user testing numbers, an article saying "users like this better" is useless.

nobody is saying. just "more".

but openrouter says they don't expect the price to change other than through the deepseek api, other people hosting the same model will keep charging the same price.

"all of them are tagged with transponders" seems like a very dangerous assumption to base a system on.

yeah, it really feels like the economics of AI are going to settle on trust - who do you trust to act on your behalf, because that's where the real value comes in.

and at the same time, it feels like all the AI companies - not just elon - are doing everything they can to burn trust.

brick pavers need a tamped gravel base, then bedding sand, and then polymeric sand to lock them together. the cost of install is not just the cost of pavers. and installation is something that requires enough skill that it's possible to do it wrong. depending on how much weight you plan to drive over it, you might also need to pour a concrete base for them.

geogrid is orders of magnitude easier and quicker to install.

sort of. they have a local model, it does some things. they also have significant cloud infrastructure backing it, and most tasks are going to be sent off to the cloud for processing, not be handled by the on-device model. Siri is not on-device by any stretch of the imagination.

yeah, i tried out GLM-5.2 when the news was all full of hype for that, and it's fine... definitely better value that API rates for claude. but comparing the value i got from that to the value i get from a claude max subscription... claude is way cheaper.

that matches my experience with their GA stuff though.

Does that mean anything at all? They released it. Customers paid for it. Sticking a preview label on it doesn’t change that it was released

>build reliable software

this feels like "make no mistakes" level of prompting. reliable software isn't as simple as making it reliable, it's about choosing the trade-offs in the areas that don't matter as much as the areas that do. if you keep prompting the LLM to make your software more robust it will keep giving you things to do. they aren't all good things. eventually you'll end up needing kubernetes to run a calculator app.

there was a significant impact on the market because the market assumed that a strong cheaper model would have a significant impact on the US AI companies' business. That didn't happen.

these questions all have easy answers that could be quite easily discovered by simply trying to use passkeys, instead of trying to find reasons not to use them.

my priority for a TV isn't privacy or user control, it's minimizing annoyance. and appletv, roku, or similar streaming boxes delivers that best.

reasonably important, if you're never connecting your lg tv to any network, and just connecting it to an AppleTV or something and letting HDMI-CEC control it so you never even see their OS.