upvote
This is just cache reads. In real usage it costs 15% more than Fable 5 -- all for marginal gains.

https://artificialanalysis.ai/

reply
Cache reads dominate in modern workflows (coding CLIs and modern web clients such as ChatGPT Work and Claude Cowork (web)).
reply
Output tokens are 5x more expensive than input tokens, so I'm not sure "dominate" is entirely correct.

A conversation with 20 turns, 50k tok growth per turn, 1m tok context at end would price out like this:

Fable 5 ($1/M cache reads) ; cache reads 9.5M tok × $1.00 = $9.50 ; cache writes 1M tok × $12.50 = $12.50 ; output 1M tok × $50 = $50.00 ; total = $72.00

Fable 5.1 ($0.25/M cache reads) ; cache reads 9.5M tok × $0.25 = $2.38 ; cache writes 1M tok × $12.50 = $12.50 ; output 1M tok × $50 = $50.00 ; total = $64.88

So yes, cheaper, but not massively.

reply
The big issue they face right now is that vastly cheaper open models are proving capable for more and more uses at cents on the dollar.

This is the right direction, but they aren't going to get there fast enough.

They will list, investors who don't know anything about tech will buy, the world will realise that China just put out a model that is good enough at a fraction of the price, they will crater.

reply
I hope that this also applies to the Subscription usage. As that can then stretch out Fable usage by a lot more.
reply
Sounds like it specifically does not apply to subscription usage.
reply
That's load bearing!
reply
But does it fail open or close?
reply
The gate is green
reply
My understanding is subscription usage generally has free cache reads, but I'm not sure if maybe Fable was different in that regard.
reply
The fact that we don't know is part of the problem. Subscription usage has always been pretty opaque.
reply