Jay Saha
3.6K posts

Jay Saha
@chotathanos
Software Engineer with the sun Understanding baremetal with the moon
Noida, India 参加日 Mart 2015
854 フォロー中306 フォロワー

@UnTalNixon_exe @precisox So which device can handle a great model?
Can DGX spark do this?
English

Pero el hecho de que el Ryzen AI Max+ tenga 128 GB compartidos no lo convierte en una supercomputadora: su ancho de banda de memoria (~500 GB/s) sigue estando muy por debajo de los más de 1,000 GB/s de una GPU de gama alta o los sistemas de Apple Silicon de gama Max/Ultra. Es ideal para modelos de 8B a 32B a toda velocidad, no para monstruos de frontera.
Español

AMD acaba de dar un golpe fuerte en la IA local.
Lisa Su subió al escenario con un mini PC del tamaño de un libro grueso en una sola mano y ejecutó en vivo un modelo de 235 mil millones de parámetros. Sin datacenter. Sin cloud. Sin alquilar GPUs.
El protagonista es el Ryzen AI Max+ 395 (Strix Halo). Es el primer chip x86 que une CPU y GPU con 128 GB de memoria unificada. En Linux, el GPU puede usar hasta ~110 GB de esa memoria.
Para ponerlo en contexto: una RTX 5090 tiene 32 GB y una 4090 tiene 24 GB. Este pequeño equipo ofrece más del triple de memoria accesible para modelos grandes, en un chasis compacto.
En pruebas específicas de inferencia (como DeepSeek R1), superó en más de 3x al rendimiento de una RTX 5080 cuando el modelo no cabe en la VRAM de la tarjeta de Nvidia.
El precio real del equipo con 128 GB (GMKtec EVO-X2) suele estar entre $1,800 y $2,500 según ofertas (el kit oficial de AMD es más caro).
Para quien usa mucho IA, esto cambia las cuentas: en vez de pagar cientos de dólares al mes en suscripciones (Claude, ChatGPT Pro, Cursor, etc.), puedes correr modelos potentes localmente con Ollama, LM Studio o similares. Privacidad total, sin límites de tokens y sin que te corten el servicio a las 3 a.m.
No es que las suscripciones vayan a desaparecer mañana, pero para muchos casos de uso (RAG con documentos privados, prototipos, agentes locales, etc.) esta opción se vuelve muy atractiva.
Estamos viendo el inicio de una nueva etapa de IA local accesible y potente??
Español

This is actually nuts. The IT department of Rio de Janeiro's city government just dropped a 397 billion parameter AI model.
And it's beating Alibaba's latest Model. 😨
It's called Rio 3.5 Open 397B. Built by IplanRIO the people who run the city's websites and digital services. A municipal IT company.
397 billion parameters. 1 million token context window. Open-source under MIT license. Strong on coding, agent tasks, and multilingual benchmarks.
Meanwhile Alibaba's Qwen 3.7 once the most loved open-source model in the world went proprietary. API only. No weights. The community moved on overnight.
And in that gap, two models stepped in:
→ MiniMax M3 : 1M context, 59% on SWE-Bench Pro, open weights coming
→ Rio 3.5 : built by a city government. Competing with frontier models
A trillion-dollar Chinese tech giant went closed. A Brazilian city hall went
open.
2026 is genuinely unhinged.
𝗭𝗲𝗻 𝗠𝗮𝗴𝗻𝗲𝘁𝘀@ZenMagnets
Alibaba Qwen3.7 slowly fading into irrelevance at the frontier due to proprietary stance. In it's place we have Minimax M3 and... *checks notes* Rio 3.5 397b, made by the municipal IT company of Rio de Janeiro's city government. huggingface.co/prefeitura-rio…
English

@AnoopKaippalli After all the praise the credi goes to modi government 😂
English

🇺🇸 US: Diesel
🇨🇦 Canada: Diesel
🇦🇺 Australia: Diesel
Major economies run double-stack freight trains, but they still rely on diesel because standard electric lines can't clear two-story trains.
But instead of settling, India custom-built a 7.5-meter high-rise grid. Today, India is the FIRST and ONLY country on Earth running double-stack containers on pure electric power. This is the scale of transformation happening under the Modi govt.
English

@GoogleResearch This is one of the coolest thing I always wanted to come alive.
Compute everywhere!!
English

Today on the blog, we discuss a pathway for the second life of phones through the exploration of “phone cluster computing”, which can directly reduce the environmental footprint of computing by avoiding the need for further raw material extraction. More →goo.gle/4aJe5vO
GIF
English

Trials of triple stack container train. Yes, it does exist.
Indian Tech & Infra@IndianTechGuide
🚨 India is the first and only country to operate double- stack container trains with electric locomotives.
English

@CourageousRo Comedy shows are ruining people's life more than anything
Its good to see even after the show the jokes are never over 🤣
English

Sejal Pawar belongs from a Reservation quota?? a ST background?? BUT HOW ??
Pawar is from the lineage of Rajput, No state of India Consider them ST then how she got a ST certificate ?
There is definitely a use of fake certificates to get admission in medical colleges with lower marks and it should be investigated.
Her degree must be cancelled If she has done any fraud in the caste certificate.


English





















