Thanks in large part to @linoy_tsaban's posts, I've created an image creation and editing suite with all the LoRAs and tools I've come across for Qwen Image 2.1
I can use it on a local computer and in a Space I’ve created so anyone can use it: huggingface.co/spaces/Maximo…
If you’d like more information on how I set it up, here’s my post where I explain it: maximofn.com/en/qwen-image-2…
MaximoFN retweeted
👋 Hola, Ministerio de Transformación Digital, @mintradigital:
👋 Hola, Secretaría de Estado, @SEtelecoGob:
Autofirma es un componente esencial de la Administración Electrónica española. Millones lo utilizamos para relacionarnos con la Administración. No por placer, sino porque es ✨requisito✨.
----8<----8<----8<----8<----8<----8<----
El paquete de Autofirma que distribuís para Mac no cumple las garantías de seguridad modernas (notarización), así que el sistema dice que es malware que puede dañar el sistema y rechaza instalarla. 😃🔫
----8<----8<----8<----8<----8<----8<----
🤦 Para instalarla hay que —ojo al dato— desactivar las políticas de ciberseguridad del sistema operativo. 🤦
¡Hola, @INCIBE! Hacéis campañas fabulosas para concienciar a la ciudadanía de los riesgos de ciberseguridad, pero luego nos enjaretáis —el Estado— marrones como este. ¿Podéis hacer algo, por favor?
Respetado ministro @oscarlopeztwit:
Respetada secretaria de Estado @mariagv:
— Si hacéis software, cumplid los estándares modernos de empaquetado y distribución de software.
— Si no podéis notarizar Autofirma, modernizad la arquitectura del programa para que se pueda.
— Si no podéis modernizar Autofirma, necesitamos un Ministerio para la Transformación del Ministerio de la Transformación Digital.
Esto es lo que ve un usuario al intentar instalar Autofirma en un Mac:
MaximoFN retweeted
Still no answer from TeamYouTube, and their DMs are closed.
The only option they've left me, instead of giving me real support and not just replying through automated bots, is to invoke EU law, specifically Articles 14, 15, 17, and 20 of the Digital Services Act and demand a detailed written explanation. I don't know if you have something similar in USA. This is not optional. It's binding EU law.
Maybe that way I can reach Google Community Outreach, an actual human who can review my case.
It's ironic, to say the least, that their systems are the ones treating me 100% through automation, while my channel is the one being accused of being automated, even though it's the innocent one!!
This is huge, because we were seriously considering, from Freepik/Magnific, creating a series of productions and pushing them hard on YouTube... but this is making me completely rethink it (although is not 100% my call, I have some weight on it).
And I doubt we're the only ones feeling this fear right now: who would want to venture into creating a fiction series in 3D or with AI on YouTube, reach millions of viewers, and then suddenly get the channel demonetized as 'inauthentic' simply because it's 'generated' (3D/AI/etc.) plus 'fiction'?
This is becoming personal for me. I can't help it.
I'm earning like 20 bucks / month on this channel 😂, LOL, but this is not about money. It's about principles, and about valuing human involvement in my creations, even when using AI tools... tools like Nano Banana and Veo 3, which are created by Google itself!
If I have to spend 50k on lawyers to get back a 20-bucks-a-month channel... I will absolutely do it, xD.
For the glory of what is RIGHT.
Hey @TeamYouTube, @YouTubeCreators!
My channel @javilopen on YouTube got demonetized (wrongly IMHO) for "Inauthentic content": content that appears mass-produced / automated.
Respectfully, I'd love to show what actually goes into each video (and if you are just randomly seeing this, it works kind of like a tutorial, because I'm showing some cool tricks in the screenshots 😉):
1/ Real human scripts and narratives
2/ Prompts crafted for hours until each image/video is "perfect"... well, perfection doesn't exist
3/ Hundreds of generations per shot / video (a video is 15 seconds maximum but each generation takes minutes), and I keep less than 5% of what I generate
4/ Strong edition (that takes several weeks for certain videos) in DaVinci Resolve / CapCut + plugins
5/ Color grading and post production
6/ Sometimes: real voice actors, foley artists (for the sounds), etc
This is literally the opposite of mass-produced. Some of the piece takes weeks/months!!!
Appeal already submitted. Hoping it reaches a human who can see the work 🙏
Reposts appreciated!
MaximoFN retweeted
Hey @TeamYouTube, @YouTubeCreators!
My channel @javilopen on YouTube got demonetized (wrongly IMHO) for "Inauthentic content": content that appears mass-produced / automated.
Respectfully, I'd love to show what actually goes into each video (and if you are just randomly seeing this, it works kind of like a tutorial, because I'm showing some cool tricks in the screenshots 😉):
1/ Real human scripts and narratives
2/ Prompts crafted for hours until each image/video is "perfect"... well, perfection doesn't exist
3/ Hundreds of generations per shot / video (a video is 15 seconds maximum but each generation takes minutes), and I keep less than 5% of what I generate
4/ Strong edition (that takes several weeks for certain videos) in DaVinci Resolve / CapCut + plugins
5/ Color grading and post production
6/ Sometimes: real voice actors, foley artists (for the sounds), etc
This is literally the opposite of mass-produced. Some of the piece takes weeks/months!!!
Appeal already submitted. Hoping it reaches a human who can see the work 🙏
Reposts appreciated!
Esta imagen explica muy bien lo que creo que va a pasar con la IA y el desarrollo de SW
Hasta ahora nos estábamos centrando en los problemas de la izquierda, porque eran asequibles para la mayoría de la población, mientras que los de la derecha muchas veces ni se abordaban
En resumen, la IA nos quita tiempo de problemas de poco valor y nos da tiempo en resolver problemas de alto valor. Cuando aprendamos a resolver esos problemas, entrenaremos a la IA en ello y nos volverá a dar tiempo para resolver problemas de mayor valor aun
MaximoFN retweeted
My basic model of capabilities: LLMs are good at problems similar to those that appear in their training data.
Training data largely reflects the world, and so LLMs are relatively good at problems that are common, relatively bad at problems that are rare.
Gráfica del uso de APIs a través de LangSmith desde diciembre de 2025
Me sorprende el poco uso de la API de OpenAI, yo esperaba que la de OpenAI fuese la de la derecha y la de Gemini la de la izquierda
Bastante de acuerdo con @juanmacias
La IA es la culpable de los despidos masivos!
Pero no solo eso, fue también la culpable del asesinato de Kennedy y fue quién mató a Manolete 😱😱😱😱😱
Anuncian a bombo y platillo que los subagentes llegan a Gemini CLI!
En los tiempos de la IA es como si se hubiesen descubierto hace 5 años y lo implementan ahora!
Subagents have arrived in Gemini CLI! 🤖🚀
Create your own custom subagents in @geminicli! Subagents are specialized, expert agents that the main agent can delegate work to.
📦- Subagents have their own set of tools, MCP servers, system instructions, and context window.
🏷️- Use @agent to explicitly delegate to a subagent
🧹- Keeps the main context window clean
⚡️- Speed up work by running agents in parallel
Read more in the launch blog below 👇
¿Recuerdas cuando te uniste a X? ¡Yo no! Me lo ha recordado X #MiAniversarioDeX
Nuevo modelo Kimi K2 supuestamente SOTA
Aunque ya no me creo mucho los benchmarks, este modelo tiene muy buena pinta, superando a GPT5 y Sonnet 4.5, por mucho menos precio ($/1M tokens input/output)
* Kimi K2: 0,15/2,50
* Sonnet 4.5: 3,00/15
* GPT5: 1,250/10,000
Ejecuta hasta 200-300 llamadas secuenciales a herramientas sin intervención humana
Así que un muy buen modelo para agentes que probaré
Pesos: huggingface.co/collections/m…
Blog: moonshotai.github.io/Kimi-K2…
Documentación: platform.moonshot.ai/docs/ov…
¿Confirmación de manipulación de benchmarks por parte de OpenAI?
Hace tiempo publiqué que OpenAI publicaba los resultados de sus LLM en el benchmark SWE-bench verified sobre 477 muestras, en vez de sobre las 500 muestras del benchmark



















