i talk about tech, AI and automation | @aiscwork
SF
Joined April 2024
- Tweets5.8K
- Following621
- Followers209K
- Likes12K
roman retweeted
AI video is going from cool demos to an actual business way faster than I expected
Jensen Huang (NVIDIA CEO) backing Higgsfield makes a lot of sense when you look at the numbers
40M+ users and 150M generations a month is pretty wild
Jensen called the founders of Higgsfield “visionary in the work that they do” and “some of the brightest people in the world”
In my opinion the interesting part is seeing AI video move beyond random clips into something people can actually use for work
Curious to see how far this goes over the next year
roman retweeted
El CEO de Anthropic viendo como China lanza un nuevo modelo de IA que supera a Claude Opus 5.5 en CASI TODO costando 70% MENOS y no puede hacer nada al respecto
Introducing Kimi K3.1 Pro: our most capable 3T-class open-weight model.
🔹 1M-token context, 2.8T total / 104B active params (MoE), Multimodal.
🔹 Full-project coding: games in Unreal, Unity & Godot.
🔹 Excels at parametric 3D modeling in Blender & CAD.
🔹 More reliable at finance work: Accurate modeling with hallucination down by 65%.
🔹 Matching the strongest proprietary models across coding, agentic and knowledge benchmarks.
Kimi K3.1 Pro and K3.1 Pro Highspeed (2–3× faster) are now live on Kimi.ai, Kimi Work, Kimi Code, and the Kimi API.
Open weights by Nov 4, 2026.
🔗 API: platform.kimi.ai
🔗 Tech blog: kimi.ai/blog/kimi-k3-1-pro
roman retweeted
NVIDIA decide invertir en Higgsfield.
Los jugadores de LoL cambiando los bailes de sus personajes por gatitos nada más enterarse:
holy f*ck... NVIDIA just invested into higgsfield
Jensen Huang said
"the founders of Higgsfield are visionary
in the work that they do… but these are
some of the brightest people in the world"
$1B in ARR
40 million users
but yeah higgs is "just another wrapper"
roman retweeted
Los CEOs de las demás empresas de vídeo con IA viendo cómo NVIDIA invierte en Higgsfield
DST Global. Goldman Sachs. Nvidia.
There’s NO big investor left who isn’t in Higgsfield.
Jensen Huang called the founders “visionary” and explained why: “you invest in people, not companies.” He also added that Higgsfield got the brightest people in the entire world.
In an industry obsessed with benchmarks and GPU counts, the smartest money keeps backing a team that builds for creators.
BY THE PEOPLE, FOR THE PEOPLE
roman retweeted
🚨 LA BURBUJA DE LA IA PUEDE EXPLOTAR INCLUSO SI LA TECNOLOGÍA FUNCIONA
Y Jensen Huang acaba de explicar una de las razones.
La demanda de nube de IA es enorme. Pero el capital escasea.
Ayer:
→ Nasdaq: −2,6 %
→ Meta: −7 %
→ Semiconductores: −6,7 %
El problema: la infraestructura se paga HOY.
Los ingresos llegan DESPUÉS.
Y necesitas financiar la diferencia.
En medio de estas dudas, Nvidia acaba de invertir en HIGGSFIELD.
La plataforma de vídeo e imagen con IA afirma superar los 1.000 millones de dólares de ingresos anualizados.
Nvidia apuesta por aplicaciones que pueden justificar la infraestructura que vende.
Más clientes generando vídeos.
Más necesidad de cómputo.
Más mercado para sus chips.
PORQUE TENER LA MEJOR TECNOLOGÍA NO BASTA SI LAS CUENTAS NO SALEN.
@jensenhuang and NVIDIA just invested in Higgsfield
We turn competition at the foundation layer into compounding advantage for our customers. Every advance at the frontier expands what Higgsfield can deliver
The moat is ownership of demand, and we clearly own one.
roman retweeted
Este ha sido el progreso de la IA en los últimos 3 años 👀
Ahora imaginad lo que viene con la inversión de NVIDIA en Higgsfield
Jensen Huang: “… the founders of Higgsfield are visionary in the work that they do … but these are some of the brightest people in the world.”
We're grateful to Jensen and the teams at NVIDIA and TCV for believing in us. Both companies have joined Higgsfield as investors in our Series B
extension.
Since our August Series B:
• We’ve crossed $1 billion in annualized revenue.
• Enterprise contracts are up 71%, growing 100%+ month-over-month
• We’ve grown from 30 million to over 40 million users globally.
TCV has backed Netflix, Facebook, ByteDance and Spotify, companies that changed how people create and consume media. NVIDIA has powered every generation of Higgsfield, from L40 and A40 GPUs through Hopper and Blackwell, with Vera Rubin next to meet growing demand.
Thank you to our creators and customers for building with us every day, and to everyone at Higgsfield whose work brought us here.
The full Bloomberg Announcement is linked below.
roman retweeted
NVIDIA has access to a front-row seat across the entire AI industry.
It sees the models, the infrastructure, and the companies building on top of them.
And it chose to invest in Higgsfield.
That tells you something about where value is being created in AI: not just in building models, but in turning them into products people actually use at scale.
@jensenhuang and NVIDIA just invested in Higgsfield
We turn competition at the foundation layer into compounding advantage for our customers. Every advance at the frontier expands what Higgsfield can deliver
The moat is ownership of demand, and we clearly own one.
roman retweeted
Jensen Huang: “… the founders of Higgsfield are visionary in the work that they do … but these are some of the brightest people in the world.”
We're grateful to Jensen and the teams at NVIDIA and TCV for believing in us. Both companies have joined Higgsfield as investors in our Series B
extension.
Since our August Series B:
• We’ve crossed $1 billion in annualized revenue.
• Enterprise contracts are up 71%, growing 100%+ month-over-month
• We’ve grown from 30 million to over 40 million users globally.
TCV has backed Netflix, Facebook, ByteDance and Spotify, companies that changed how people create and consume media. NVIDIA has powered every generation of Higgsfield, from L40 and A40 GPUs through Hopper and Blackwell, with Vera Rubin next to meet growing demand.
Thank you to our creators and customers for building with us every day, and to everyone at Higgsfield whose work brought us here.
The full Bloomberg Announcement is linked below.
Higgsfield, an AI video startup founded by a former Snap executive, said it has attracted investment from Nvidia and TCV to help fund global infrastructure and hiring to capture growing demand. bloomberg.com/news/articles/…
roman retweeted
Kimi K3.1 acaba de cargarse a Opus 5.5 / Astra
Le di a Kimi K3.1 el mismo prompt, y creó esta escena por 5 veces menos tokens que Opus 5.5
Kimi K3.1: $2.15 · 6 mins
Opus 5.5: $10.65 · 28 mins
China acaba de volver a la carrera de la IA
Introducing Kimi K3.1 Pro: our most capable 3T-class open-weight model.
🔹 1M-token context, 2.8T total / 104B active params (MoE), Multimodal.
🔹 Full-project coding: games in Unreal, Unity & Godot.
🔹 Excels at parametric 3D modeling in Blender & CAD.
🔹 More reliable at finance work: Accurate modeling with hallucination down by 65%.
🔹 Matching the strongest proprietary models across coding, agentic and knowledge benchmarks.
Kimi K3.1 Pro and K3.1 Pro Highspeed (2–3× faster) are now live on Kimi.ai, Kimi Work, Kimi Code, and the Kimi API.
Open weights by Nov 4, 2026.
🔗 API: platform.kimi.ai
🔗 Tech blog: kimi.ai/blog/kimi-k3-1-pro
roman retweeted
🚨 ESTO ES UNA LOCURA
Un proyecto de GitHub que convierte tu ordenador en un agente de IA que sigue trabajando aunque cierres la terminal.
Se llama Mercury Agent y es open source.
¿Qué puedes hacer? 👇
→ Darle acceso a tus archivos y ejecutar tareas
→ Hablar con el mismo agente desde Telegram, Discord o Slack
→ Recordar tus preferencias y proyectos entre conversaciones
→ Crear agentes especializados que trabajen por su cuenta
→ Automatizar tareas y programarlas para más tarde
→ Controlar los permisos y limitar el gasto en tokens
Lo interesante es que no es otro chatbot que simplemente responde preguntas.
Puede investigar, escribir código, trabajar con GitHub y ejecutar tareas en segundo plano.
Además, puedes elegir su personalidad, los modelos que utiliza y qué acciones puede realizar.
Ideal para quienes quieren tener su propio agente de IA funcionando 24/7.
Y sí, puedes consultar el código en GitHub.
Enlace abajo 👇
🚨 ESTO ES UNA LOCURA
Un proyecto de GitHub que convierte conceptos técnicos en máquinas interactivas creadas con IA.
Se llama Anatomy y es open source.
¿Qué puedes hacer? 👇
→ Crear ilustraciones isométricas con un nivel de detalle brutal
→ Explicar sistemas complejos con objetos que representan cómo funcionan
→ Añadir piezas móviles e interacciones
→ Generar gráficos SVG con efectos WebGL y elementos 3D
→ Crear visualizaciones para documentación, webs y presentaciones
Lo interesante es que no genera el típico diagrama lleno de cajas y flechas.
La IA construye una representación visual del concepto, con piezas que puedes explorar e incluso mover.
Ideal para explicar ideas técnicas de una forma mucho más visual.
Y sí, puedes consultar el código y probar los ejemplos en GitHub.
Enlace abajo 👇
roman retweeted
Claude Code es un lío. Hasta que instalas esto.
Hay un plugin oficial de Anthropic llamado claude-code-setup.
Te dice qué automatizaciones puedes montar (hooks, skills, MCP servers, subagentes…) y cómo configurarlas paso a paso.
Básicamente analiza tu proyecto y te recomienda qué activar.
Para instalarlo:
/plugin install claude-code-setup@claude-plugins-official
Guarda este post para no perderlo 🔖
roman retweeted
AI made everything faster. Love is the one thing it shouldn't rush.
Introducing Inyo: your personal matchmaker that lives in iMessage.
No feed to scroll. Nothing to swipe. You just text.
Inyo gets to know you the way a close friend does, including the patterns you can't see yourself. Then, every few days, it introduces you to someone who fits.
Two separate lives. One introduction.
We're live in NYC, SF, LA, and Toronto. Comment "Inyo" and I'll send you an invite.
roman retweeted
We started Mubit in Aberdeen. No network, no lab, nobody knew us. We just wanted AI systems that actually work.
Today we work with some of the best companies and engineers in the world.
And the thing we kept seeing: you and your competitor run the same AI. It should be yours. 🧵
Comment "SAPLING" in the next 48 hours for $100 free credits. Enterprises get the first month free.
roman retweeted
Un ingeniero de Anthropic acaba de revelar cómo convertir Opus 5.5 en el modelo más barato del planeta.
Pásale este prompt a Opus 5.5, ya me lo agradecerás.
Después, lee la guía de abajo sobre las funciones de Claude que casi nadie conoce.
Guárdate esto, te va a servir.
roman retweeted
SIGAN A ESTA PERSONA NO DIGO NADA MÁS...
Introducing P-Video-2-Pro-Avatar, based on @MiniMax_AI H3 !
Turn a single portrait into a realistic talking video with natural expressions and accurate lip sync.
And for launch, it's 50% off for the first two weeks, until November 5 at 18:00 CET!
• 𝗣𝗿𝗶𝗰𝗶𝗻𝗴: $0.02/s at 480p · $0.04/s at 768p
• 𝗦𝗽𝗲𝗲𝗱: ~0.97 s at 480p · ~2.2 s at 768p per second of output video
• 𝗜𝗻𝗽𝘂𝘁: One portrait with a text script or uploaded audio
• 𝗩𝗼𝗶𝗰𝗲𝘀: 30 voices across 10 languages
• 𝗗𝘂𝗿𝗮𝘁𝗶𝗼𝗻: 1s–3min at 24 fps, in 480p or 768p
• 𝗖𝗼𝗻𝘁𝗿𝗼𝗹: Voice delivery, scene motion, seed, and prompt upsampling
One image, two ways to bring it to life. Use your own audio or generate speech directly from text.
Available via our inference partners @replicate, @eachlabs, @Cloudflare, @scenario_gg, @wavespeed_ai, @WiroAI, Sociaro, @TellersAI, and @togethercompute.
📚 Model docs: docs.api.pruna.ai/guides/mod…
🧩 Try the model through the API or Playground: dashboard.pruna.ai/login
roman retweeted
the weirdest part of buying training data is having to commit before you really know what you're getting
weeks of outreach, NDAs and quality checks just to find out a dataset doesn't fit
@DataVendor_ai lets labs inspect the evidence first and sample tasksets in their own HUD team before buying
this should've existed years ago
We're announcing the public beta of datavendor.ai
We spent the last two years building the best solution for people buying and selling data, whether you're a researcher in a lab or a company trying to sell data.
Data is the fundamental bottleneck for AI progress, and DataVendor will be the infra that solves it.
roman retweeted
why should accessing AI models come with an extra markup?
sume API is taking a different approach: image and video models at their original public API prices, with a familiar UI and MCP support for Claude and ChatGPT.
the real value should be in what you build with AI, not extra layers of markup.
try Sume.com today and save 50% on AI video costs!
As a builder in the AI video space, I was terrified about how much markup API providers charge for simply routing the models.
Higgsfield API adds a 40% markup even with its maximum advertised discount. Fal adds 100%. Runway adds 414% (Seedance 2 480p)
I might be hated by the industry, but :
Today, we’re launching Sume API, the first image & video model provider with 0% markup.
For example with Seedance 2.5 1080p
• 50% cheaper than fal
• 50% cheaper than higgsfield api (29% even with biggest discount)
• 28.02% cheaper than kie ai
The routing infrastructure layer is never the moat. We believe that APIs should be commoditized, and the moats will be built at the application layer.
Check the pricing comparisons below, and see how the broken industry is saved.
roman retweeted
Introducing P-Video-2-Pro-Avatar, based on @MiniMax_AI H3 !
Turn a single portrait into a realistic talking video with natural expressions and accurate lip sync.
And for launch, it's 50% off for the first two weeks, until November 5 at 18:00 CET!
• 𝗣𝗿𝗶𝗰𝗶𝗻𝗴: $0.02/s at 480p · $0.04/s at 768p
• 𝗦𝗽𝗲𝗲𝗱: ~0.97 s at 480p · ~2.2 s at 768p per second of output video
• 𝗜𝗻𝗽𝘂𝘁: One portrait with a text script or uploaded audio
• 𝗩𝗼𝗶𝗰𝗲𝘀: 30 voices across 10 languages
• 𝗗𝘂𝗿𝗮𝘁𝗶𝗼𝗻: 1s–3min at 24 fps, in 480p or 768p
• 𝗖𝗼𝗻𝘁𝗿𝗼𝗹: Voice delivery, scene motion, seed, and prompt upsampling
One image, two ways to bring it to life. Use your own audio or generate speech directly from text.
Available via our inference partners @replicate, @eachlabs, @Cloudflare, @scenario_gg, @wavespeed_ai, @WiroAI, Sociaro, @TellersAI, and @togethercompute.
📚 Model docs: docs.api.pruna.ai/guides/mod…
🧩 Try the model through the API or Playground: dashboard.pruna.ai/login
roman retweeted
Metí 250$.
Este bot los convirtió en +13.000$.
Y ni siquiera lo programé yo.
Cogí un proyecto open source de GitHub.
Le dediqué un par de horas en Claude Code para limpiarlo, entender cómo funcionaba y hacer algunos ajustes.
Le di a ejecutar.
Pensaba que iba a fundir la cuenta en menos de una hora.
Minimicé la terminal y seguí con mi día.
Cuando volví por la noche, ya no eran 250$.
Eran +600$.
Desde ese momento, simplemente dejo los logs abiertos en el segundo monitor y veo cómo ejecuta operación tras operación.
La clave está en el timing.
El bot espera hasta los últimos dos minutos de cada ronda de 5 minutos de BTC.
Para entonces, Bitcoin ya ha hecho gran parte del movimiento y la dirección suele estar bastante definida.
En ese momento entra a favor de la tendencia, comprando contratos que cotizan entre 0,80$ y 0,99$.
Unos minutos después, si el movimiento se mantiene, la posición se liquida a 1$.
La lógica es sorprendentemente simple:
→ Esperar hasta que falten unos 2 minutos para el cierre de la ronda
→ Comprobar que BTC ya se ha movido entre 70$ y 100$ durante ese intervalo
→ Operar siempre a favor del movimiento, nunca en contra
→ Cubrir una pequeña parte de la posición únicamente si el mercado se desequilibra demasiado (por ejemplo, 95/5)
Es como comprar un boleto de lotería cuando prácticamente ya han anunciado el resultado.
El código es público y cualquiera puede usarlo en su pc:
github.com/radioman/polymark…
Gratis.
El bot fue creado siguiendo el tutorial de abajo, échale un ojo si quieres crear el tuyo👇