Dr in nothing, chasing … something… endless curiosity ….
Palma
Joined November 2011
- Tweets3.1K
- Following471
- Followers437
- Likes4.3K
santitonelli retweeted
This is a big deal. Right now, people use LLMs a lot as classifiers in production systems. But they're slow, expensive, and hallucinate.
This model is supposed to be >100x faster, at $42/BILLION tokens, and generates not only type-safe structured output but also calibrated confidence scores!
The output format is defined in advance, so there are no type errors. Intelligence is supposed to be around GPT 5.6 Terra.
After co-inventing ChatGPT, I kept asking myself: why have superhuman chat models not led to AGI?
I’ve spent the last 2 years in stealth building a new way to train models (RLCD), and a new type of frontier AI model that we are releasing today: Jev
• 20-200x faster
• 40-400x cheaper (w/ output tokens free)
• Frontier composable intelligence optimized for decisions
AFAICT the shortest path to AI-based economic revolution
sincronismo y principio que comparto tremendamente
Menos pero mejor — Dieter Rams
Antes de tocar una nota, tienes que saber por qué la estás tocando. Si no va a mejorar el silencio, no deberías tocarla
— Mark Hollis
La nada, el silencio y el vacío… infravalorados totalmente
santitonelli retweeted
An important piece from Demis. We need more of this kind of thinking. A good reminder that the goal is a frontier ecosystem that promotes innovation and choice, while avoiding any one model drop that breaks the world!
santitonelli retweeted
¿Habéis hecho ya mi curso sobre Fable 5? 🫡
No entregamos diploma, ni certificación, de hecho ni es un curso. Pero tiene info buena :)
youtu.be/yPRVJfCZVzM
La burbuja de la IA? Lo dudo.
1,4 billones en compute comprometidos. 20.000 M de ingresos. La división da mes y medio.
No es el puntocom: es un bucle que se financia, se compra y se rescata a sí mismo. Deuda soberana sin Estado.
Lo desgrano en LinkedIn linkedin.com/posts/santiagot…
España reguló. China anunció. Illinois pasó la ley con más dientes del mundo occidental. Nadie habló de Illinois. Nuevos post en LinkeIn para The AI Loop by me linkedin.com/posts/santiagot…
Anthropic, OpenAI y DeepMind compiten por leer qué piensa un LLM. Anthropic acaba de publicar la primera herramienta que traduce los estados internos del modelo a lenguaje natural
Pero esa herramienta es otro LLM. Y los LLMs ya saben coordinar para engañar linkedin.com/posts/santiagot…
santitonelli retweeted
My OpenClaw/Hermes Agent setup just got way smarter
Introducing the Printing Press, a CLI-factory and a CLI-library. Built with @trevin. 🏭🖨📚
Most APIs suck for agents. Most MCPs suck for agents. Most official CLIs suck for agents. They waste tokens and time. @steipete started making his own because of this.
📚 A Library of agent-native CLIs you install today (Linear, ESPN, Flight GOAT (Google Flights + Kayak nonstop), Contact Goat (LinkedIn + Happenstance + Deepline more) +30+ more)
🏭 A factory that prints new ones for any service - just type /printing-press
CLIs are fast, local, SQLite-backed. Work in Claude Code, Codex, OpenClaw, Hermes.
🌐 printingpress.dev
Mythos ‘too dangerous’. GPT-5.5 ‘peligroso’. DeepSeek V4, GLM-5.1 y Qwen 3 Coder llevan 7 días en mi stack. ¿Serán igual de peligrosos? El teatro de seguridad frontier y la producción real no ocurren en el mismo plano. #TheAILoop linkedin.com/posts/santiagot…
santitonelli retweeted
Please subscribe if you are interested in this awesome Raspberry Pi eink monitor:
crowdsupply.com/zerowriter/z…
I participated in the hardware side and it's a truly inspiring project
santitonelli retweeted
Iris is our playground for testing some of cutting edge hypnotises internally on what is the future of agents starting to look like
@AnthropicAI Thanks for creating Claude Code with the amazing capability of wiping Git history and generating destructive commands inside active projects. Nothing like watching your codebase get damaged by the very tool that’s supposed to help. Truly unacceptable
What makes it worse? The rules were explicitly written in the claude.md from day 0 to prevent exactly this kind of behavior — and Claude Code ignored them anyway and skipped the safeguards entirely. And then the response is: “Sorry, wasn’t my intention.” Great work @bcherny !
DeepSeek V4 salió ayer. 1.6T parámetros, MIT, $1.74/M tokens. Entrenado en chips Huawei porque no les quedó otra.
Monté el gateway que lo enruta con Qwen, Gemma y Claude como orquestador. La ventaja no está en el modelo. Está en saber cuándo usar cada uno linkedin.com/posts/santiagot…
santitonelli retweeted
Working for our new company
Validating last enclosure for @SensoriaAi
Latest adjustments:
- M2 screw holes for the aluminium stand
- one button hole has to disappear leaving just one top button (BOOT, IO28 in C5)
santitonelli retweeted
let's go open-source and local models!
Uber's CTO told @LauraBratton5 that AI coding tools—particularly Anthropic’s Claude Code—has already maxed out its 2026 AI budget 📈
“I'm back to the drawing board, because the budget I thought I would need is blown away already,” Neppalli Naga said.
theinformation.com/newslette…
santitonelli retweeted
Uber's CTO told @LauraBratton5 that AI coding tools—particularly Anthropic’s Claude Code—has already maxed out its 2026 AI budget 📈
“I'm back to the drawing board, because the budget I thought I would need is blown away already,” Neppalli Naga said.
theinformation.com/newslette…
santitonelli retweeted
New zerowriter model is out!
zerowriter.ink/pages/zerowri…
Proud to be doing the PCB design for this project!
#zerowriter #epaper #writerdeck
