What I cannot understand is that the cost per 1M of tokens keeps going down.
This stuff is converging on nearly free, especially if you don’t use the most expensive model and even more especially if you opt to host your own infrastructure and run models yourself.
As someone who doesn’t know much about their paid tiers, if you host it on your hardware, are there really some that are worth paying for vs just using one that’s available for free? I kind of assumed there wasn’t much of a difference and that’s why they were trying to ban ones who had “outside influence.”. Because in the end they’ll all be using mostly the same data sets and the differences will be who is manipulating the answers to influence the end users, which has some profit to it, but not nearly as much as they would need
I’m not an expert either. I mostly use Opus at work and I don’t pay for it - the company does. I’ve noticed Fable almost seems like a waste for a lot of tasks, so mostly I don’t use it. I’ve not tried dialing it down to Sonnet or even Haiku to see how things fare, but keep meaning to try that out.
You’ll see Fable 5.2 is at the very top with 61.1. Even Opus is 58.7.
Here, you can see the very top is 54.9 from Kimi and I’m not sure which of these you can host yourself. My understanding is that just because something is open weight doesn’t mean you can download it and in some instances, even when you can, the hardware to run it is not something exactly consumer-class (but a company might choose to build or rent that hardware themselves).
When you remember how people went gaga over this in 2022…imagine using a model that “lags” some of the top-tier stuff but applied to more mundane things - it’s not like the capabilities from 2022 or whatever evaporated. Suppose you assume you can run something at decent enough speeds, but it “lags” the others by 6 months to a year?
At some point, it just seems to me there will be diminishing returns to keep paying for the most expensive option. My company offers me Fable now and I don’t find that I need it for day to day things. For all I know, Haiku might be enough for some things.
It’s a problem they’re running headlong into. Anthropic basically is going to suck all the money out of the market and their IPO suggests they’re going on a spending spree and will issue more debt, which makes it even tighter.
It’s going to be an absolute shitshow if it’s anything at all like dot-com bubble. Having things lose 80% of their value or go to zero even may be something edgelords think is exciting and/or hilarious, but the fallout is something all of us will feel.
Where are they raking the money from? Who is giving them billions of dollar? Are all investors and CEOs dumb as a bucket?
I’m so confused.
debt, wallstreet etal buying their bonds
What I cannot understand is that the cost per 1M of tokens keeps going down.
This stuff is converging on nearly free, especially if you don’t use the most expensive model and even more especially if you opt to host your own infrastructure and run models yourself.
As someone who doesn’t know much about their paid tiers, if you host it on your hardware, are there really some that are worth paying for vs just using one that’s available for free? I kind of assumed there wasn’t much of a difference and that’s why they were trying to ban ones who had “outside influence.”. Because in the end they’ll all be using mostly the same data sets and the differences will be who is manipulating the answers to influence the end users, which has some profit to it, but not nearly as much as they would need
I’m not an expert either. I mostly use Opus at work and I don’t pay for it - the company does. I’ve noticed Fable almost seems like a waste for a lot of tasks, so mostly I don’t use it. I’ve not tried dialing it down to Sonnet or even Haiku to see how things fare, but keep meaning to try that out.
If you look at leaderboards like:
https://llm-stats.com/leaderboards/llm-leaderboard
You’ll see Fable 5.2 is at the very top with 61.1. Even Opus is 58.7.
Here, you can see the very top is 54.9 from Kimi and I’m not sure which of these you can host yourself. My understanding is that just because something is open weight doesn’t mean you can download it and in some instances, even when you can, the hardware to run it is not something exactly consumer-class (but a company might choose to build or rent that hardware themselves).
https://llm-stats.com/leaderboards/open-llm-leaderboard
When you remember how people went gaga over this in 2022…imagine using a model that “lags” some of the top-tier stuff but applied to more mundane things - it’s not like the capabilities from 2022 or whatever evaporated. Suppose you assume you can run something at decent enough speeds, but it “lags” the others by 6 months to a year?
At some point, it just seems to me there will be diminishing returns to keep paying for the most expensive option. My company offers me Fable now and I don’t find that I need it for day to day things. For all I know, Haiku might be enough for some things.
It’s a problem they’re running headlong into. Anthropic basically is going to suck all the money out of the market and their IPO suggests they’re going on a spending spree and will issue more debt, which makes it even tighter.
I was looking at their own info on their IPO, and their profit is a negative $23.3 billion.
https://ipo.anthropic.com/us/
Bubble will be even more glorious when it bursts
Some of us have lived through bubbles bursting, and it’s never fun for us normal folks. You might want to temper your expectations a little bit.
Yes, I too have lived through many of those bubbles. Did it suck for me? Yes. Do I still enjoy seeing my enemies fail? Yes.
I can have it both ways even if you cannot.
I probably can because I’m most likely not in danger of losing my job and apartment. This time.
That can’t be said for everyone, and I guess I’ve got enough empathy to give a shit about that
You have empathy for billionaires?
No I have empathy for the people who’s lives are going to be fucked when this bubble pops.
Decidedly not the billionaires. They will, again, be just fine.
In fact, it’ll be a fire sale for them and many of them will get more wealthy.
It will suck for a lot of the people, but the bubble has to pop before things can even start to improve.
It’s going to be an absolute shitshow if it’s anything at all like dot-com bubble. Having things lose 80% of their value or go to zero even may be something edgelords think is exciting and/or hilarious, but the fallout is something all of us will feel.
I don’t think it will be at all glorious.
Well, it’s a lot bigger than dot com. Concerning.
I can enjoy the failings of billionaires while being concerned for the common people at the same time.
The billionaires are going to be basically unaffected
That’s exactly the thing.
They will live and have food but they don’t care about that. They care about money and they’ll lose billions.
In order of appearance: futures investment, other AI companies, yes.