Madridista with BRAIN 🧠

1+1=2

[🖤™ @Ilvaite0]

Mourinho made him play the exact role Rodri would play in the new system and he did quite good, maybe there’s a chance for redemption?

[Kerry @tea_with_hunny]

A follow up on Imu being a parrot: ngl this guy cooked. I genuinely love this.
#ONEPIECE

[Rub✧ @Rubyrz_]

One Piece fans came up with a theory that Imu is a parrot and it’s actually pretty funny when you think about it So Roger laughed at Laugh Tale because he found out the ruler of the world was actually a parrot then Roger said “It’s such a funny story” fck lol😭😭

[SHANKS @reddhairshanks]

Part 2 of Imu is a parrot theory

[Claude @claudeai]

New in Claude Cowork: teach Claude a skill.

Record your screen while you do a task, talk through it as you go, and Claude turns it into a skill it can run again. Find it under Record a skill in the + menu of the Claude desktop app.

Available on Pro, Max, and Team plans.

[Fiction @0xF1ction]

Kimi CEO Zhilin Yang:

"Claude thinks more data is the edge - it's not - the data's already gone

the real edge is wasting less of it - 2x the intelligence from every token - that's Kimi 3"

his one big idea: token efficiency is the new Moore's Law - twice the intelligence, same data, no extra GPU

- he threw out Adam, the optimizer every lab's run since 2014. 15 trillion tokens, not one spike

- a curve he called "the most beautiful thing I've seen in 2025"

the real goal? "every token a model generates is non-fungible" - a CEO's mind isn't a designer's. taste is the frontier nobody's building for

watch & bookmark

[Linas Beliūnas @linasbeliunas]

Instead of watching 2 hours of Netflix tonight, watch this in-depth interview with Kimi founder Zhilin Yang.

He explains why everyone’s racing on reasoning while Claude quietly bet on agents - but the real hard part isn’t the agent itself, it’s the model underneath.

His memorable metaphor: a pure reasoning model is like “a brain in a fish tank” - it can think forever but never touches the world. An agent is that brain wired to tools, memory, and action.

Yang shares how they’re thinking about K2 helping build K3 through strong agent skills, the biggest bottlenecks in generalization, test-time scaling in two dimensions, and even how RL-style thinking applies to both models and running teams.

Whether you’re building agents, scaling models, or just want to understand frontier AI thinking beyond the usual Western narratives, this is one of the clearest and most thoughtful long-form discussions available right now.

Bookmark it & watch - it might be the most valuable 100 minutes you spend on real AI engineering and strategy this month.

[sunick @Serantych]

Kimi's CEO Zhilin Yang:

"OpenAI, Google, everyone just kept scaling data. that

game is over - we ran out of internet."

here's the shift he's betting on:

→ scaling data hit a wall - the high-quality internet is basically used up

→ O1 opened a new axis: scale the thinking at inference, not just the training data

→ his bet: the training-to-reasoning ratio flips, and that's where the next startups win

bookmark this ↓

[Biggies Malls 2.0 @Biggies_MaIIs]

Trump when he handed out the runners up medal to Simeone

[Kenpachi @Kenpachi1070]

Je lance la propagande Kylian Mbappé ballon d’or

Montez dans le train

[Red @Beelal_Red]

@ik_ohene7 Bath of corruption

[Megatron @Megatron_ron]

BREAKING: 🇺🇸🇨🇳 China is now leading the AI ​​race after spending ten times less of what the US did

Kimi-K3 by @Kimi_Moonshot is now number 1 in the Frontend Code Arena with 1679 pts, surpassing Claude Fable 5.

[Pøgba Senior @TheSaItIsHere]

Those Tuchel decisions make sense when you find out he has family in Argentina that migrated there around the 1940s

[Michael Sikand @michaelsikand]

China's communist roots really showing up with IPO valuations that are actually attractive.

[KawzInvests @KawzInvests]

It's honestly disgusting how openly retail gets farmed here.

I'm not bearish these companies. I just think the valuations US retail gets handed on day one are ridiculous.

All the reasonable-entry upside already happened in private, walled off from us. By the time we're allowed in, the tag reads:

- SpaceX, ~$1.8T at IPO
- Anthropic, ~$1T, still private
- OpenAI, ~$850B, still private
- Cerebras, ~$80B on day one

Now look at where Chinese retail gets to step in:

- Zhipu, ~$7B at IPO
- MiniMax, ~$6B at IPO
- CXMT, ~$80B with the move still ahead, and the public can actually subscribe

American retail only gets access once the valuation is already in the stratosphere.

The 100x always happens where you're not allowed to be. You only get invited when it's time to hold the bag.

[هَيا @hhayya__03]

الحياة بعد إقصاء إسبانيا والأرجنتين

[Nick Cruse 🥋 @SocialistMMA]

Russia realizing the entire world has no problem with them killing Lindsey Graham

[14. @0xwinniffy]

jude bellingham vs. norway ( motm ) https://t.co/KNbSAaoEO1