byzantinegene

8 ms·

1.
by byzantinegene 6d ago
22x more to be exact
2.
by byzantinegene 6d ago
i think the main consensus here is that the actual performance is not indicative of the benchmark performance (which supposedly outperforms the previous iterations)
3.
by byzantinegene 6d ago
i'm not sure why there's a need for comparison here. both are not saints and you shouldn't trust any of them anyway
4.
by byzantinegene 23d ago
I think the (non-definitive) evidence is that the chinese models are always just slightly behind the publically released US models, and never ahead.
5.
by byzantinegene 23d ago
China is still quite far from monopolizing hardware production
6.
by byzantinegene 24d ago
most of the bubble right now is fueled from circular financing of Anthropic and OpenAI
7.
by byzantinegene 28d ago
100b sounds implausible if you consider that anthropic's revenue is only expected to reach 100b this year, 100m is more likely.
8.
by byzantinegene 1mo ago
where are the cheaper chinese models?
9.
by byzantinegene 1mo ago
source? most benchmarks show "smart" routing producing near-identical accuracy and 50-75% cheaper costs.
10.
by byzantinegene 1mo ago
i think plenty of people don't want to be forced to using just Claude and ChatGPT
11.
by byzantinegene 1mo ago
they have no reason to do that to harm their reputation if their goal is profit, in which in this case it is.
12.
by byzantinegene 1mo ago
this would never work if the core business model requires your flagship product to be best-in-class.
13.
by byzantinegene 2mo ago
such a business plan has not yet created economic value, being able to roll out features at rocket speed is not a huge determinant of a startup's success.
14.
by byzantinegene 2mo ago
sam and dario would be extremely happy
15.
by byzantinegene 2mo ago
why would you even hire 2 AI friendly people when you could just prompt it yourself? we wouldn't even need you anymore
16.
by byzantinegene 2mo ago
we are already at the point where AI proponents are planning to have the review cycle be replaced by llms
17.
by byzantinegene 2mo ago
i doubt the users really care as long as they get the job done, AI in 90% of use cases is just a race to the bottom, nobody cares as long as it's cheap and gets the job done.
18.
by byzantinegene 2mo ago
fraud that cannot be reliably be prevented, which makes the business model in-question fall into 'questionable' category.
19.
by byzantinegene 2mo ago
good, then the money could be potentially put to better use
20.
by byzantinegene 2mo ago
The official reasoning they use is sanctioning 'intellectual theft' due to distillation.
21.
by byzantinegene 2mo ago
VCs have been horseshit for some time and throwing money at horseshit startups to increase their AUM, their incentive structures are fundamentally misaligned with their clients.
22.
by byzantinegene 2mo ago
Why would model progress slowing be necessarily bad for China?
23.
by byzantinegene 2mo ago
if this is true, the frontier labs are not able to justify their trillion dollar valuations, they are barely making anything on subsidized plans.
24.
by byzantinegene 2mo ago
I believe they will last until they IPO, and not long after that. $200/mo plans are not good for their P&L when their users using $10000 worth api credits. That's -98% margin loss per user.
25.
by byzantinegene 2mo ago
until they put their money where their mouth is
26.
by byzantinegene 2mo ago
claude code is free to use, you can use claude code with any non-anthropic model, so harness stickiness doesn't benefit the AI labs at all.
27.
by byzantinegene 2mo ago
the reselling of fixed price plans by resellers are causing american labs to lose alot of money and that's why they're trying really hard to stamp it down.
28.
by byzantinegene 2mo ago
it's not like the protests are working anyway
29.
by byzantinegene 2mo ago
does demolishing homes to build data centers count?
30.
by byzantinegene 2mo ago
guess that's what AI trained itself on

More ›