🐈‍⬛ᯓ★🐈

Si llegaste hasta acá es porque de seguro te domé, zurdo drogadicto.

[🐈‍⬛ᯓ★🐈 @MarchioParello]

@kick @vitalyzdtv Lol you mfs have no shame

[Lior Alexander @LiorOnAI]

You can now run 70B LLMs on a 4GB GPU.

AirLLM just made massive models usable on low-memory hardware.

𝗪𝗵𝗮𝘁 𝗷𝘂𝘀𝘁 𝗵𝗮𝗽𝗽𝗲𝗻𝗲𝗱
AirLLM released memory-optimized inference for large language models.

It runs 70B models on 4GB VRAM.
It can even run 405B Llama 3.1 on 8GB VRAM.

𝗛𝗼𝘄 𝗶𝘁 𝘄𝗼𝗿𝗸𝘀
AirLLM loads models one layer at a time.

Instead of loading everything:
→ Load a layer
→ Run computation
→ Free memory
→ Load the next layer

This keeps GPU memory usage extremely low.

𝗞𝗲𝘆 𝗱𝗲𝘁𝗮𝗶𝗹𝘀
• No quantization required by default
• Optional 4-bit or 8-bit weight compression
• Same API as Hugging Face Transformers
• Supports CPU and GPU inference
• Works on Linux and macOS Apple Silicon

𝗪𝗵𝗮𝘁 𝘆𝗼𝘂 𝗰𝗮𝗻 𝗱𝗼
• Run Llama, Qwen, Mistral, Mixtral locally
• Test large models without cloud GPUs
• Prototype agents on cheap hardware

[🐈‍⬛ᯓ★🐈 @MarchioParello]

@Janekeithduran Sobre todo 😂

[🐈‍⬛ᯓ★🐈 @MarchioParello]

@irefari @cr_madrista Igual, cómo vas a comparar el cementazo con las causas que tiene abiertas Chaves que son solo eso y no han pasado de ahí... Agradezca la calidad de presidente que tiene.

[Branko @brankopetric00]

The microservices conversation nobody wants to have:

Your 8-person team now maintains:
- 23 services
- 23 CI/CD pipelines
- 23 sets of dependencies to update
- 23 potential points of failure
- 1 person who understands how it all connects

That person is on vacation.

Amazon can do microservices because they have 10,000 engineers.
You have 8.

You're not building Netflix architecture.
You're building job security for the one person who set this up.

A well-structured monolith with clear boundaries beats a microservices mess every single time.

[🐈‍⬛ᯓ★🐈 @MarchioParello]

@rincidium @Remotion What's PR?

[Hector Rodriguez (Riggs) @RiggsCR]

Si usted no sabe que pasa en la CCSS, lo invito a ver este material.
Da miedo la corrupción que hay allí, encima de eso Álvaro Ramos se puso a jugar de vivo siendo el idiota que es.
Que relajo de verdad

[George Noble @gnoble79]

OPENAI IS FALLING APART IN REAL TIME

I've watched companies implode for decades.

This one has all the warning signs.

OpenAI declared "Code Red" in December.

Altman sent an internal memo telling employees to drop everything because Google's Gemini 3 is eating their lunch. Salesforce CEO Marc Benioff publicly ditched ChatGPT for Gemini after using it for two hours.

ChatGPT traffic fell in November. Second month-over-month decline of 2025. Meanwhile Gemini jumped to 650 million monthly active users.

The company that was supposed to build AGI can't keep its chatbot competitive.

But the real story is the money...

OpenAI lost $12 BILLION in a single quarter according to Microsoft's own fiscal disclosures.

Deutsche Bank estimates $143 billion in cumulative negative cash flow before the company turns profitable.

Their analysts put it bluntly: "No startup in history has operated with losses on anything approaching this scale."

They're burning $15 million per day on Sora alone.

$5 billion annually to generate copyright-infringing memes.

Even Sora's lead engineer admitted the "economics are currently completely unsustainable."

Here's the big math problem nobody wants to discuss:

It's going to cost 5x the energy and money to make these models 2x better.

The low-hanging fruit is gone.

Every incremental improvement now requires exponentially more compute, more data centers, more power.

Reports suggest OpenAI's large training runs in 2025 failed to produce models better than prior versions.

GPT-5 launched to widespread disappointment. Users called it "underwhelming" and "horrible." OpenAI had to restore GPT-4o within 24 hours because users preferred the old model.

Altman had promised GPT-5 would make GPT-4 feel "mildly embarrassing." Instead, users complained it was worse at basic math and geography.

They've released GPT-5.1, GPT-5.2 since.

Same complaints each time: too corporate, too safe, robotic, boring.

The talent exodus makes this even worse:

CTO Mira Murati. Gone.
Chief Research Officer Bob McGrew. Gone.
Chief Scientist Ilya Sutskever. Gone.
President Greg Brockman. Gone.

Half the AI safety team departed. Multiple executives reportedly cited "psychological abuse" under Altman's leadership.

And now Elon Musk is suing for up to $134 billion.

A federal judge just ruled the case goes to jury trial in April. There's "plenty of evidence" that OpenAI's leaders promised to maintain the nonprofit structure that Musk funded.

Musk provided $38 million in early funding based on those assurances. Now he wants his share of the $500 billion valuation.

OpenAI called it "harassment." But the judge disagreed.

Here's what I think happens next:

The AI hype cycle is peaking.

The diminishing returns are becoming impossible to hide.

Competitors are catching up.

The lawsuits are piling up.

OpenAI needs to generate $200 billion in annual revenue by 2030 to justify their projections.

That's 15x growth in five years while costs keep exploding.

Even Sam Altman admitted investors are "overexcited" about AI.

His exact words: "Someone is going to lose a phenomenal amount of money."

If I were running an AI startup with good traction right now, I'd be looking for an exit. Sell into the hype before the music stops.

My positioning:

I'm not touching OpenAI-adjacent plays at these valuations. The risk profile is astronomical.

If you're exposed to the Magnificent 7 through AI infrastructure bets, consider trimming. The gap between promised revolution and delivered reality has never been wider.

The smart money is rotating into sectors where valuations actually reflect fundamentals.

Small and mid-caps are trading near decade lows relative to Big Tech while earnings growth is only marginally lower.

Markets can price risk. But they can't price chaos.

And OpenAI is chaos dressed up in a $500 billion valuation.

[🐈‍⬛ᯓ★🐈 @MarchioParello]

@dsmith1713 @Darth_Seti No seas tan idiota para hablar. El BCR es una m��quina innecesaria en nuestro país.

[🐈‍⬛ᯓ★🐈 @MarchioParello]

@dsmith1713 Ojalá privaticen la caja y el BCR 👍🏼

[🐈‍⬛ᯓ★🐈 @MarchioParello]

@armandomayorga Son la oposición y son políticos. Allá usted que les cree.

[🐈‍⬛ᯓ★🐈 @MarchioParello]

@DefendamosCR @UHN_Plus Cállate maldito hijo de perra anti patria.

[🐈‍⬛ᯓ★🐈 @MarchioParello]

@DEsceptikon @UHN_Plus Sos empleado del estado no me digas 😂

[🐈‍⬛ᯓ★🐈 @MarchioParello]

@EdelZorro @UHN_Plus Para los que son de afuera: Este tal Barba Roja es un zurdo resentido, de baja escolaridad que lo único que desea es que nosotros los trabajadores de la clase productora/privada, le paguemos sus subsidios y vivir del estado. Ya con eso quedas eliminado, payaso.

[UHN Plus @UHN_Plus]

🇨����‼️ | ALERTA INTERNACIONAL — Laura Fernández, candidata presidencial de Costa Rica y la mejor alineada con Washington, denunció una ofensiva coordinada, hipócrita y desesperada de la vieja clase política y medios tradicionales para frenar el cambio iniciado con Rodrigo Chaves.

En un comunicado publicado ayer domingo, Fernández fue tajante al marcar distancia de corruptos, pactos oscuros y agendas comunistas, advirtiendo que el ataque simultáneo confirma que su proyecto rompe privilegios y amenaza al club de siempre.

[🐈‍⬛ᯓ★🐈 @MarchioParello]

@Janekeithduran @davidso16529877 De hecho tener esa foto le da aún más propiedad para hablar.

En cambio usted, dándole cancha a esta rata asquerosa de Ariel Robles.

[🐈‍⬛ᯓ★🐈 @MarchioParello]

@Alo_Aloc @silvanasc25 Se están dejando llevar por un majijo y les da ternura nada más. Pero recuerden que es del PLN. No se les puede olvidar eso. Es una rata igual que todos los pericos.

[Wall Street Apes @WallStreetApes]

Barack Obama's ICE chief gets an award for 920,000 removals

Donald Trump’s ICE chief gets called a Nazi

It’s the same person, Tom Homan

In 2012 Obama deported 409,000 people
In 2025 Trump deported 290,000 people

THE DIFFERENCE IS THE MAINSTREAM MEDIA BRAINWASHING

[Elon Musk @elonmusk]

Same person