juantomas (mAGIa is coming ) retweeted
Acabo de probar la última actualización de TensorFold.dev 1.1.1
Qwen3.5-2B Q/4 bits ->95,08 tok/s
Tiempo hasta el primer token
88 ms
La cosa no sería impresionante si no fuera porque es una RTX 3050 Vram 4Gb
Gracias @ashxhart
Ahora me toca ayudar a otros -> localIA
Replying to @robertosantosx2
This is awesome, thanks for testing it! Love seeing what people can do with a 4GB card.
juantomas (mAGIa is coming ) retweeted
1/ En abril me uní a Cursor para trabajar en Origin, su ‘git forge’, un poco antes de que SpaceXAI lo adquiriera. Cuando terminé en Unity/Plastic SCM en 2023 pensaba que no iba a volver a desarrollar un control de versiones,
We’re excited to be day 0 partners for @reflection_ai’s first open-weight model! For a limited time,
Beam is FREE on Ollama! In partnership with @reflection_ai this model is hosted in the US with full ZDR.
It has a 1M token context window and is designed for software engineering, reasoning, and long-running agentic tasks.
Run Beam on Ollama’s Cloud:
ollama run beam
Claude Code
ollama launch claude --model beam:cloud
Codex
ollama launch codex --model beam:cloud
OpenCode
ollama launch opencode --model beam:cloud
Free to use for a limited time only.
Beam is now available across platforms, with open weights under the Apache 2.0 license - in collaboration with an ecosystem of 40+ partners.
We're releasing:
- Weights, tokenizer, and reference implementation
- BF16, FP8 and NVFP4 weights and a DFlash2 drafter
- The open source harness we built and trained Beam on: Mirror CLI
Weights and model card on Hugging Face: huggingface.co/reflection
juantomas (mAGIa is coming ) retweeted
herdr-projects v1 is out - rewritten from scratch as a mission control for @herdrdev 🐏
heads up: it's a breaking change. want the old version? it lives on in the legacy branch and the v0.2.35 tag.
how v1 works:
→ talk to one coordinator - say what needs doing, it turns it into tasks on a board
→ press tab and a task gets a worker agent - claude, codex or pi (more coming)
→ each worker builds in its own worktree and branch, kept up to date with main
→ when it's done, your checks run and a second agent reviews it - fails go straight back to the worker
→ the board shows what needs you, what's ready to approve and what's still working
→ you can jump in to steer agents in herdr anytime
→ board is available on your phone via herdr-web-ui
also in: automations (schedule and event based), best-of-n, time and token limits, fallback agents, and much more
install: herdr plugin install eliasstravik/herdr-projects
on 0.2? run the same install - your old projects are set aside in ~/.herdr-projects-legacy, never deleted
repo: github.com/eliasstravik/herd…
enjoy!
juantomas (mAGIa is coming ) retweeted
Replying to @eliasstravik @herdrdev
will try this. constantly trying to manage which of my agents actually needs something from me
juantomas (mAGIa is coming ) retweeted
Replying to @ian_finlay @herdrdev
amen same here - please lmk if you run into any issues, happy to help and fix any bugs asap!
juantomas (mAGIa is coming ) retweeted
Introducing gemma.ai
The new hub for Gemma, including
📚Inference and training guides
🧠 Deep visual architectural guides
🖼️ Online demos, even running models in the browser
And a lot more to come, check it out!
juantomas (mAGIa is coming ) retweeted
Jev was multimodal all along
Yep, the Jev you know and love was an image model this whole time. We just chose to disable it because models tend to be less intelligent on other modalities.
Header: TypeSafe-Danger-Zone-Images: true
juantomas (mAGIa is coming ) retweeted
Holy Smokes! It's 3 dimensional! I did not expect this!!!
Gemini 4 Argon is here.
To show you what it can do, here are 5 builds from our team 🧵 ↓ xitter.cf/fofrAI/status/21100430…
juantomas (mAGIa is coming ) retweeted
Gemini 4 Argon is here.
It’s available globally today for developers and enterprises in the paid tier of the Gemini API and the Gemini platform, for Ultra subscribers in Google Antigravity, and Ultra and Pro subscribers in the Gemini app and Search.
Hope you all enjoy it!
juantomas (mAGIa is coming ) retweeted
I've been running Gemini 4 for a week and it is comparable with Fable in the benchmarks I tried, website migrations, at a lower price. Good stuff
Argon has this amazing natural understanding of the world. It can do so many things from first principles.
Here's a pelican riding a bike, using fourier epicycles.
Video made by Argon.
Time-synced Lyria 3.5 music prompted by Argon.
juantomas (mAGIa is coming ) retweeted
Replying to @fofrAI @Heaney555
Wooooah motion, time synced, freaking 3D Fourier whaaaaa?
Fourier is one of my favs
juantomas (mAGIa is coming ) retweeted
Hace unos días que quería escribir esto. No es una receta exacta, pero a mí me ha funcionado estupendamente bien.
Sobre cómo hacer las cosas con pasión, disfrutando y ayudando a los demás. Y cómo todo vuelve
También en la nueva web borjaperezfrances.com/corazo… View article
juantomas (mAGIa is coming ) retweeted
La cantidad de automatización que vamos a poder tener aquí va a ser loca.
Ya estoy visualizando agentes invocados para ejecutar tareas específicas vía API, CLI, Webhooks…
Esto es el comienzo de algo muy TOCHO.
Efectivamente, hoy oficialmente tenemos los workspaces disponibles para toda la comunidad.
Nuestros workspaces podrán tener pre-configurados (de forma opcional):
- Hermes
- Pi
- Gentle-Shell
- Herdr
Este último es importante porque los workspaces estan pensados NO SOLO para que corran agentes 24/7. Sino para trabajar en remoto desde cualquier dispositivo.
Lanzas una sesión en herdr de tu agente favorito y podrás acceder a ella desde el ordenador, desde el móvil o desde donde quieras. Todo corre sobre nuestra infra para que puedas tener tu workspace siempre disponible.
Como en esta sesión de ejemplo. Estoy ejecutando Gentle-Shell en mi workspace y controlando todo desde mi Mac y mi iPhone 🔥
Tenéis doc de todo aquí:
nan.builders/docs/workspaces
juantomas (mAGIa is coming ) retweeted
Love Diogo’s vibe
Reminds me of a bygone era of geeky, enthusiastic builders.
launched a model
accidentally served trillions of tokens a day
29.4% of the fortune 500 showed up
raised a really big series A from @a16z
it's been 3 weeks
we would like to sleep now






























