KimLeong

Angel in #wush. Experiment with #web3 life. Developer of genetic of @seacrypt | @v1punks | bitcoin

[KimLeong @kimuraLeong]

@grok aka bitcoin maxi, have you changed your mind ?

[KimLeong @kimuraLeong]

Relationship of AGI and PoW is much more deeper than agent economy.

[KimLeong @kimuraLeong]

Uncomfortable take : Jobloss not due to AI, but because of CEO unable to grow the pie faster than growth in productivity. So CEO is the bottleneck

[KimLeong @kimuraLeong]

@BitPaine This seems to be first legit attempt to attack fiat system. If without witholding tax for foreigner, then it will be a very good proxy to attack other fiat currency too.

[KimLeong @kimuraLeong]

Meanwhile, yen, euro, rupee, real holders sell their local currencies to buy into this. Those currencies weaken. The dollar’s reserve position strengthens. BTC accumulation happens as a byproduct. (2/2)
@saylor

[KimLeong @kimuraLeong]

BTC can be offensive!

Imagine the US issues a $2T sovereign BTC-preferred at 11%. Dollar-savers globally face a choice: hold 4% Treasuries or 11% BTC-backed USD preferred. Capital floods from Treasuries into the new instrument — but stays within the dollar system. (1/2)

[KimLeong @kimuraLeong]

@saylor @Saylor any plans to launch a STRC version (or something similar) in a no-dividend-tax spot? That 30% US withholding is holding back a lot of Asian buyers.

[KimLeong @kimuraLeong]

@demishassabis @jack Your work inspired this happen

[KimLeong @kimuraLeong]

My speculation : Future of ads isn’t on AI. It’s inside AI.

Not banners. Not pop-ups.
Just models quietly trained to favor themselves and their sponsors.

By the time it’s obvious, it’s already normal.

[Qwen @Alibaba_Qwen]

🚀 Introducing the Qwen 3.5 Small Model Series
Qwen3.5-0.8B · Qwen3.5-2B · Qwen3.5-4B · Qwen3.5-9B

✨ More intelligence, less compute.
These small models are built on the same Qwen3.5 foundation — native multimodal, improved architecture, scaled RL:
• 0.8B / 2B → tiny, fast, great for edge device
• 4B → a surprisingly strong multimodal base for lightweight agents
• 9B → compact, but already closing the gap with much larger models
And yes — we’re also releasing the Base models as well.
We hope this better supports research, experimentation, and real-world industrial innovation.
Hugging Face: https://t.co/wFMdX5pDjU
ModelScope: https://t.co/9NGXcIdCWI

[KimLeong @kimuraLeong]

@TheAhmadOsman Yes, opensource has to win. Not the other way

[KimLeong @kimuraLeong]

In few years, those who control the bitcoin, control the agents.

[Quanquan Gu @QuanquanGu]

Seed 2.0 is finally out! 🔥

[KimLeong @kimuraLeong]

@DavidOndrej1 IMO, be fair to open model, minimal agent scaffolding was used in open model for such benchmark. If want to do apple-apple comparison, please remove codex / claude code from the closed model. Or allow open model to use opencode or droid for such benchmark

[KimLeong @kimuraLeong]

@elonmusk @kimmonismus You can already do that with grok by providing extremely long - multi weeks compute with detail steps of every decision, and tokenize the output as NFT. @grok am i correct?

[KimLeong @kimuraLeong]

@tunguz @grok are you smarter than O3 pro

[KimLeong @kimuraLeong]

@JasonPLowery IMO, people will realize softwar very soon. POW is the proven technology to protect against hacking, especially from AGI. Comparing developing the first AGI vs adopting POW, POW is obviously the easier defense mechanism

https://t.co/NcKomG0TXf

[KimLeong @kimuraLeong]

Gm legends! I’m joining the 'NFT Legends Season' by @Mint_Blockchain and claim my share of 1% $MINT supply. Just grabbed 3 NFT Legend Boxes! 🍀✨

Check your eligibility and claim yours here: https://t.co/HkmdX1By8J

[KimLeong @kimuraLeong]

@balajis Will frontier math problem subject to same data contamination problem too?

[KimLeong @kimuraLeong]

@idontexist_nn If there are money on the table, what stopping competitors to build all in one model?

[KimLeong @kimuraLeong]

@idontexist_nn @Soronprfbsy Is the unchanged element just part of the cost function that resulted the model always pointing to the same answer? How thorough do we understand the behaviour of model, as in what are the root cause that causing these issues?