There are five people left in crypto.
But there are surprisingly ten different people doing podcasts.
Distribution channel: same (Telegram)
92
you guys are shocked why coins going down?
nigga, we worshipped @blknoiz06 for years now.
1
105
hahahahahaha.
same kols and vcs praising ansem today for reviving @solana trenches will abuse him for doing the same in next few months.
bookmark.
96
if crypto, blockchain or whatever shit this is wanna survive
it needs to go through a solid restart.
remove the larpers, salary farmers, grifters.
until then lets enjoy the blood bath.
I know a handful of folks in crypto who openly admit that
1. they are salary farming xyz chain because the market is bad.
2. They are trying the same old thing that has been proven to be ineffective, but yet doing it because itβs something they always wanted to do.
85
I know a handful of folks in crypto who openly admit that
1. they are salary farming xyz chain because the market is bad.
2. They are trying the same old thing that has been proven to be ineffective, but yet doing it because itβs something they always wanted to do.
crypto is down & you should not be shocked
- literally building ponzi & quick money extracting tools for half a decade
- people hiring their friends to help them salary farm & not based on merit
- more paul graham & socrates readers than people who knows how to get things done
jeet/acc (π¨πΏ,π³πΎββοΈ) retweeted
most people, even engineers, don't need frontier intelligence for their day-to-day work. a model from a couple months ago is usually enough.
frontier models will keep improving and hence, frontier tokens will always stay expensive. that's why llm routing felt like the obvious fix to me: pay for the frontier only when the task actually needs it.
upon digging into the state of llm routing, i found out that none of the routers can provably say they're better than using cheaper models.
on @ValsAI Router index, most gateway routers still lose on score + cost to just picking a single cheap model like deepseek v4.1 flash. and even newer stateful routers tracking session history, tool errors, and prompt caches still haven't proven they reliably beat a well-chosen default.
classifying difficulty isn't enough. you have to predict whether a stronger model will actually change the outcome, after the added cost of cache penalties.
i spent a week digging into how different routers work, what signals matter, what doesn't - formulated hypotheses and experiments to find better algorithms - but nothing seems to show promising results yet.
my bet: we don't solve general routing, we just accelerate past it. models get good enough that hardcoding a cheap 2-gen-old model handles 95% of real work, and you only reach for the frontier when you actually hit a wall.
stay away from market for atleast 2-3 months.
remember there was a huge hack & chigga needs to fill that hole.
chiggatardio.
148
crypto is literally downbad but this chinese wallet drainer kol @CookerFlips is shilling another low cap garbage to his follower.
if you are buying this shit you will have aids.
1
184
Smarty Alien Kol, who has been shilling bottom-tier garbage for the past five years, can't even recognise a parody page.
keep botting your posts.
1
158
any update? @CryptoWizardd
I personaly think #btc bottom is here at 67.5 k
Next up >>>>
82/85 k
1
326















