Joined May 2019
- Tweets 27,141
- Following 834
- Followers 3,672
- Likes 95,335
4,173 Photos and videos
Pinned Tweet
Adopting crypto's technology is unreasonable, but all progress depends on the unreasonable man.
There's nothing groundbreaking in history that ever looked reasonable while in development, and commercial viability has always taken decades of thankless work.
It's a miracle that this industry can take advantage of monkey pictures and coins about farts to plow through those decades of hardship.
11
1
28
9,751
got vertigo reading this
Jun 12
The largest USDC transfer in history.
~$4.4B sent to the Coinbase Hyperliquid deployer.
3
15
3,303
João Mendonça retweeted
Jun 12
It launches with two open-source projects from @triton_one:
Superbank — Solana's entire 400B transaction ledger, queryable in milliseconds. Up to 38x faster historical reads.
blog.triton.one/introducing-…
1
2
42
1,700
Jun 12
today all around the world tens of thousands of engineers and founders didn't have to ask for a single thing from nasdaq
they just sent transactions
3
8
675
Jun 12
so many words to say crypto's bloomberg
Jun 12
Jason Yanowitz says the Blockworks IPO is inevitable:
"Onchain businesses scale 100x. Traditional capital markets go on-chain, BlackRock, Robinhood, Stripe. Both become Blockworks customers."
"Two-sided platform: issuers provide info, asset managers consume it. AI agentic workflows on top, disclosures, IR, compliance, security. Full enterprise platform. Blockworks goes public. Vision changed when we chose to consolidate."
Onchain native. TradFi migrating. Both customers. IPO destination.
2
261
Jun 12
given the right context, you never need anything else for codegen other than v4 flash
Jun 12
I'm so enamored with deepseek v4 flash that I might actually convert to jinpingism
1
98
Jun 12
I'm so enamored with deepseek v4 flash that I might actually convert to jinpingism
161
João Mendonça retweeted
Jun 12
117
72
780
148,090
Jun 12
ngl this is pretty plausible, and aligns with the tingle that the ponzi has to survive at least til the end of this business cycle
Pio reveals the real reason why Saylor sold 32 Bitcoin
"People ask why he sold the 32 Bitcoin, and it's two reasons. They want S&P inclusion, and they want a better rating from the agencies that rate these preferred equities. Right now their rating is junk, the agencies are literally writing the Bitcoin to zero, not recognizing it as an asset."
"This sounds insane but go try it. If you have 5 million dollars of real Bitcoin on chain and you walk into a bank for a mortgage loan and offer it as collateral, they go, so that's zero. I'm not kidding. But if you hold IBIT, the Bitcoin ETF, that's totally fine, because it's just an equity."
"The S&P gatekeepers specifically said that because Saylor had demonstrated he'd never sell any Bitcoin, they couldn't view it as a viable part of his balance sheet. So he needs to show he can pay the dividend with Bitcoin if he has to. Once stretch gets a AAA rating, do you know how much capital a thing paying 11% is going to attract?"
1
1
637
João Mendonça retweeted
Jun 12
even for "planning"
when you "know" the process of building sufficient context for the implementation will not involve a lot of precision wrt code analysis, there are drastic diminishing returns on using bigger models
this is especially true if the smaller model you're using is effective at compartmentalizing code analysis such that it can go through all relevant context without needing too much breadth in the codebase
if the model can simply plow through the codebase in tiny steps and gather all the information it needs, there's no difference in accuracy to "SOTA" models
SOTA is actually a pretty bad acronym to describe "the latest models", because most of them are just bulkier, they're not "new art"
the smaller models are SOTA, they're just smaller or more optimized
1
1
430
Jun 12
look at ts
people are conditioned to looking at models as "intelligence", and consequentially smaller models are "dumber" 🙄
they do not understand what an LLM is, and if they ever did, anthropic & co is making sure their brains get properly washed
Jun 11
Why would you do that? You work for an $80B company, why impose these arbitrary limits on your ability to deliver better results?
2
3
677
Jun 11
trillions
wen fable @jinping??
OpenCode Go is becoming the best source of data on what models are being used and how
we've made a public stats page so you can see the latest
opencode.ai/data
1
643
Jun 11
how is MiMo burried under other?...
people are sleeping on xiaomi's engineers
1
257
Jun 11
the portuguese translation for "Mimo" is treat, which fits the model like a glove
o MiMo é um mimo
163
João Mendonça retweeted
Jun 11
To give you a concrete sense of what we're doing with @Backpack Securities.
- Trade stocks on Backpack
- Deposit/Withdraw to Solana
- Trade on DeFi in your wallet
- On/off ramp to your brokerage account
Tokenized stocks that are redeemable for the real thing just like a stablecoin.
Demo:
Jun 11
$SPCX goes live on @Solana via @SunriseDeFi following SpaceX’s Nasdaq debut.
Issued by Backpack Securities. On/off-ramp SpaceX shares between traditional brokers and Solana.
115
210
749
225,836
Jun 10
taiki is the biggest mid curver I've seen get idolized on here
no offense with offense
"fat protocol thesis is dead"
SOL was trading at <20PR before all the trump bs with a single breakout app
fat protocol thesis will win, we just haven't had any serious contenders
4
17
1,228
Jun 10
but he's also extremely honest transparent and humble
which is very rare in idols
so there's that
135