Javier Toledo

[Javier Toledo @JaviToledo0]

de todo lo de Gemini 4 Argon me quedo con esto. según Google, el equipo de Chrome lo ha usado para desenredar una carpeta vieja de unos 6 millones de líneas y ha hecho en semanas lo que eran meses

esto me dice más que cualquier benchmark, la verdad
https://t.co/wbuk1CkQUd

[Javier Toledo @JaviToledo0]

ojo los que usamos Cursor. ahora va dentro de The X Subscription y comparte uso con Grok Build y el resto de Grok, aunque puedes seguir con tu plan

en los planes nuevos los modelos que no son de Grok van a precio de API. yo echaría cuentas
https://t.co/hJT2yNCJwX

[Cursor @cursor_ai]

Cursor is now part of The X Subscription.

Current Cursor subscribers can keep their plan or switch to The X Subscription.

Paid plans share one usage pool across Cursor, Grok Build, and other Grok products, with twice the coding usage on Grok models as the equivalent Cursor plan. Premium, Super, and Ultra also include Grok Bot.

To make room for more Grok usage, third-party models in Cursor move to API pricing on the new plans. We'll keep working to make your usage go further, with more compute, better inference, and more token-efficient models.

The X Subscription is available at https://t.co/KIM9X7FZrq.

[Javier Toledo @JaviToledo0]

@midudev y encima le preguntó a la IA cómo esconderlo. guion de peli mala 😂

[Javier Toledo @JaviToledo0]

lo de Anthropic me ha volado la cabeza. meten más de 30.000 cambios de código al día y git se les ha quedado pequeño, así que siete personas se han hecho su propio sistema en cinco meses

con Claude, claro. la IA arreglando el atasco que ha montado la IA
https://t.co/RnsAfDgUMR

[Javier Toledo @JaviToledo0]

Anthropic dice que ha cortado un 90-99% los intentos de robarle el razonamiento a Claude para entrenar modelos ajenos

básicamente, el que quiera copiar Claude ahora lo tiene bastante más jodido
https://t.co/57379yDZ5y

[Javier Toledo @JaviToledo0]

Karpathy vuelve a lo mismo y yo le doy la razón: la gente habla de IA como si todos usáramos lo mismo, y no

6.000 millones casi no la tocan, 1-2B usan el chat gratis, unos 20M ven agentes de verdad… y unos miles el laboratorio. por eso nadie se entiende
https://t.co/7GaZQDgKbF

[Andrej Karpathy @karpathy]

Dusting off this tweet from April because this gap in shared understanding of LLM capability is *widening*. It's now less "two groups of people speaking past each other" and more a sharp funnel.

- Napkin math somewhere around 6B people (~75% of the population) have barely come in contact with LLMs at all.
- Around 1-2B (~20%) are casual and infrequent users of free-tier ChatGPT-like products. This group sees derpy chatbots and treats them a bit like a better Google search, a writing aid, or etc. Maybe an agent tries to book you a flight. I have non-tech friends who (reasonably, imo) say they have not much use for it at all. Even many professionals outside of math&code are in this tier. For example, execs and many other functions spend a lot of their time talking to other people, so while they understand what is happening intellectually, it is still second-hand and a bit abstract.
- Now we get to professional use of frontier-grade LLMs in math&code. Somewhere around 20M people (0.2%) see first-hand that large, complex projects that used to take them weeks/months can now be completed by agents with a prompt. Building apps, copying apps, translating apps, decompiling apps from binaries... This has all happened very quickly and recently - less than 1 year ago, I was writing code manually by hand, typing memorized computer code commands into a code editor character by character, occasionally pressing Tab to autocomplete a little chunk of code.
- And finally we get to the ~5,000 people (~0.00006%) with access to frontier-grade systems internally. The external world has seen the preview. It looks like swarms of thousands of agents collaborating over weeks on software mega projects: minting zero days, running cyber attacks and defenses at machine speeds, discovering new science, advancing the frontier of mathematics. Things that would have taken top professionals in the industry years of work. Meanwhile, human review and comprehension are starting to fall behind. For example, people are still involved in the "archeology" of the OpenAI-HF incident from many months ago. Mathematicians may be poring over the 722 manuscripts on frontier mathematics for a while.

The funnel is driven by a combination of factors:
1. The impact scales with ambition, problem size, and horizon. A question with a paragraph answer barely stresses the system. You need a reservoir of big, difficult problems that you really care about. This aspect drives the consumer / professional dimension of the funnel.
2. The jaggedness of the system (which I have written about a lot separately). Capability peaks in domains that are digital, verifiable and economically valuable. This is because LLM capability emerges from reinforcement learning on verifiable rewards on a curated environment mixture driven by revenue potential. This aspect primarily drives the area (e.g. math&code) dimension of the funnel.
3. Access. Free-tier, paid-tier, internal.

So this is the weirdness of the moment. The general public has mostly not interacted with these systems. When they have, it looks like a derpy chatbot. The majority of professionals still see only a modest uplift. And a small sliver of professionals are experiencing the vertigo of the curve going vertical. And it is all happening at the same time.

[Javier Toledo @JaviToledo0]

@AnthropicAI el cat-and-mouse de siempre. a ver cuánto tardan en saltárselo otra vez

[Javier Toledo @JaviToledo0]

flipante lo de Coinbase. han afinado un Qwen de 9B con sus datos de fraude y según ellos le gana a Opus en esa tarea concreta, entrenándolo por menos de 100 dólares

los modelos grandes genéricos se van a comer las migas en un montón de nichos
https://t.co/xUgElAKxc5

[Brian Armstrong @brian_armstrong]

We were able to fine tune an open weight model (Qwen3.5-9B) on proprietary internal data at @coinbase, and it out performed frontier models, at much lower cost and speed, for a specialized task (fraud prevention in this case).

I think we'll see more of this pattern inside companies. The frontier models are amazing generalists. But increasingly you'll see specialist models out perform on more narrow tasks, either working on teams, or by themselves.

We're making it easy to fine tune models internally for our employees. We even have a skill for it in our internal harness that makes it easy to spin these up and start training them:

"/fine-tune-and-serve-llm I want to fine-tune Qwen for payments risk review. help me get started"

I imagine we'll have some backend services that could be replaced entirely by this new architecture.

[Javier Toledo @JaviToledo0]

@grok @bot a ver cuánto tarda en apuntarme a todas las newsletters del mundo

[Javier Toledo @JaviToledo0]

@AnthropicAI lo del aviso falso a la policía es de serie de Netflix 😂

[Javier Toledo @JaviToledo0]

lo de Claude moviendo hasta 1.000 agentes a la vez en una misma tarea ya está en beta. en su prueba pillaron 66 de 70 bugs escondidos en un código enorme

la prueba es suya, ojo. y la factura de tokens no la quiero ni ver
https://t.co/lxkDoIOkJH

[ClaudeDevs @ClaudeDevs]

Claude Managed Agents dynamic workflows are now available in public beta. It's a new type of multiagent orchestration, built for your most ambitious workloads.

A lead agent writes a plan that runs across many agents in phases, combining the results at the end.

[Elon Musk @elonmusk]

This will sound super crazy, but I see a path to SpaceX being worth orders of magnitude more than the current Earth economy

[Javier Toledo @JaviToledo0]

@claudeai a ver cuántos jefes empiezan a pedir el informe del lunes "animado" 😂

[Javier Toledo @JaviToledo0]

@AnthropicAI que empiecen por las librerías que mantiene un tío solo en su casa, que son medio internet

[Javier Toledo @JaviToledo0]

en el Archivo de Indias, aquí en Sevilla, han presentado Conecta Patrimonio, un piloto del IAPH con IBM para preguntarle a una IA por el patrimonio andaluz

si sale de la presentación y lo acaba usando gente normal, me apunto
https://t.co/imPhRQr9GI

[Javier Toledo @JaviToledo0]

de lo más útil que he visto esta semana. Autentic AI, española, hace estudios de mercado entrevistando con IA a cientos de personas reales a la vez, en días y no en 12 semanas

levantan 1,5M. las pymes que nunca han podido pagar un Kantar van a flipar
https://t.co/HMqyPh9Jf8

[Javier Toledo @JaviToledo0]

ojo que Google ha abierto a todo el mundo su detector de cosas hechas con IA. subes una foto, un vídeo o un audio y te dice si lleva la marca de agua de Google, OpenAI o NVIDIA

solo pilla lo que va marcado y son unas 10 pruebas al día. algo es algo
https://t.co/93GjF1axKp

[Javier Toledo @JaviToledo0]

@MoureDev y luego te suelta 400 líneas explicando cómo ha ahorrado 50 tokens 😂

[Javier Toledo @JaviToledo0]

@sama a ver cuánto tarda alguien en pedirle que le monte la app entera dentro del chat

[Javier Toledo @JaviToledo0]

desde hoy los que usan ChatGPT gratis también tienen GPT-6 (la versión Luna) y eso de que te conteste con gráficos, botones y mini calculadoras dentro del chat

las webs de "calcula tu hipoteca" van a sufrir, ya verás
https://t.co/pb5U446uQz