Dani

@Jaisiero @DotCSV Ahí estás cambiando la conversación; estamos hablando de veo 3, no de genie 💀💀💀

@1aF0k4 @DotCSV x.com/i/status/20182…

Es heavy lo fuerte que fue el golpe sobre la mesa de Seedance 2.0 que paró por completo la cadencia de modelos de vídeos que llevábamos hasta entonces. En paz descansen todos esos checkpoints de modelos que fueron entrenados y nunca publicados por no estar a la altura 😪

@Jaisiero @DotCSV Veo 3 alucina de manera más evidente, no sigas defendiendo lo indefendible.

@1aF0k4 @DotCSV No pusiste la escena de la puerta. El modelo sigue alucinando. Creíble no significa plausible.

@Jaisiero @DotCSV La capacidad de Seedance 2.0 para entender físicas es extremadamente superior a Veo 3, además de la consistencia de escenas y personajes en el mismo video. Con un prompt en Veo no te acercas ni de broma a Seedance 2.0. Es buen modelo, pero ya quedó atrás x.com/charliebcurran…

@1aF0k4 @DotCSV Elabora porfa

@Jaisiero @DotCSV "Mucho más impresionante" 🥀🥀🥀

@DotCSV Veo3 sigue siendo mucho más impresionante. Generación en tiempo real que entiende el input del usuario y aparentemente simula físicas. "We are getting there" Modelos que entienden el mundo.

@zbouboee De cabrones no ser un pinche tibio

Calle 24 se he proclamado ante las controversias que se han desencadenado hoy durante Junior H en la mañanera junto a la presidenta de México Claudia Sheinbaum.

nah my classmates really dont know about gpt 5.5 they're still all on 5.3 instant and complaining that it's shit

@gkisokay What's the actual best setup for 16gb vram + 64gb ram?

The Local LLM cheat sheet for your 16GB RAM device I pulled together a lineup of small models that can run comfortably on a Mac Mini or personal laptop while still leaving room for context without melting your machine. Models for Daily Use Qwen3.5 9B / GGUF / Q4_K_M Daily driver. General chat, drafting, research, translation. If you're keeping only one, keep this. DeepSeek-R1 Distill Qwen 7B / GGUF / Q4_K_M Reasoning engine. Math, logic, step-by-step problems. Slower, but worth it when you need actual thinking. Models for Specialty Work Qwen2.5 Coder 7B / GGUF / Q4_K_M Code specialist. Completions, refactors, debugging, repo Q&A. Better than a generalist when the task is code. Llama 3.1 8B / GGUF / Q4_K_M Long context worker. RAG, doc chat, codebase Q and A. The output isn't top tier, but the context is strong for its size. Phi-4 Mini Reasoning / GGUF / Q4_K_M Compact thinker. Logic, structured answers, math, and short coding bursts. Smaller context is the catch. Models for Efficiency Gemma 4 E4B / GGUF / Q4_K_M Light all-rounder. Writing, chat, light agents, structured output. Phi-3.5 Mini / GGUF / Q5_K_M Pocket sidekick. Summaries, extraction, background doc chat. Easy to pair with a bigger model. Qwen3.5 2B / GGUF / Q4_K_M Useful for summaries, tagging, rewrites, and lightweight sidekick work. Micro Models Qwen3.5 0.8B / GGUF / Q5_K_M Classification, keyword routing, binary decisions, triage. Gemma 4 E2B-it / GGUF / Q4_K_M Lightweight chat, quick Q and A, summaries, tiny agents. My personal choice for a single model is Qwen3.5 9B For two models use Qwen3.5 9B + Qwen2.5 Coder 7B for code, or Qwen3.5 9B + Phi-3.5 Mini for support tasks. Let me know in the comments your experience with these models, or any I have left out.

@outsource_ Also Depseek V4 drop this week… again😶‍🌫️

What a Monday so far... 🤯Lovable LEAK 🤖Kimi 2.6 👥Kimi Agent 🚨Qwen 3.6 MAX Preview 🚀GPT 5.5 Spud Releasing SOON 😉Herems-Workspace V2

keep scrolling bro im sure the next post will be life changing

@JoelDeTeves How do you connect it to marketplace? I read that meta has a lot of rules about automatic scrapping and shit like that. Are you using like a browser agent?

Changed my Hermes Agent streaming token to a paperclip btw

@LottoLabs .gguf

@LottoLabs I can’t wait for UD-Gemma-4-30B Claude Opus 4.6 Distill (Q4_M) 😭

Please god be a dense 20-30b model

@TeksEdge @OpenRouter I don’t know if it’s my configuration but I spent 1.5m tokens only to open the browser and do a few clicks on a page with Qwen 3.6. I ran out of credits in openrouter 😪

Qwen3.6 wasn’t on my April bingo card but it just appeared on @OpenRouter! Testing with my 📊 benchmark and 🦞 Clawdbots.

@0x0SojalSec Do you recommend it to be a personal agent? I have a 5060ti gpu to, I’m using openclaw but I’m paying for APIs (deepseek 3.2 / gpt 5.4) and I would really like to don’t pay to scrap information and stuff like that

$500 GPU Beats Claude Sonnet on Coding. A frozen 14B model (Qwen3) running on a $500 RTX 5060 Ti GPU that beats Claude 4.5 Sonnet on LiveCodeBench. Without any fine-tuning, fully local. Trades latency for cost. 🙌

@capy_eddy @leonpalafox “Desarrollos de LLMs internos en el país”, usar una API china barata o directamente usar cualquier modelo tipo Llama 3 8b que son realmente estupidos hoy en día 🥀🥀🥀

@leonpalafox Ya no te entendí. Quieres desarrollos de LLMs internos en el país ? O no ? O solo cuando sean desarrollados por los privados (que a ti te gustan), o no ? Perfiles como el tuyo son los primeros que deberían de dar de volteretas de felicidad con esas noticias.

Lo genial es que todo mundo le va a preguntar a ChatGPT antes que a esta herramienta

Y esto que fue?

never ask a man his token usage

@CodyAlanAllred @kshvbgde Also the original price it’s $800, only with discount it’s close to NEO. I’m not an apple fan but you need to admit that neo is a good product

@1aF0k4 @kshvbgde For the store only. Shows how popular of a laptop it is. If NEO was out of stock you could apply the same talking points.

wake me up when a $599 Windows laptop with 8gb ram can do this