Angel Sirlopu Cumpa

Arquitecto de software / Developer
Papá de Thiago Nicolás

[Angel Sirlopu Cumpa @AnguelSC]

En el 2007 pedía prestada una PC estudiar a un familiar, en el 2010 mi madre hizo el esfuerzo de comprarme una PC que me duró muchos años, ahora tengo la oportunidad de gozar de un hermoso equipo y TODAS estas etapas tiene un factor en común. El apoyo incondicional de mi madre

[KC @karanC_12]

Opus 5.5 vs DeepSeek V4.1 Flash-1017

•Opus: 20 mins, $6
•DeepSeek: 1 hour, $0.23

Opus cooked, but DeepSeek got surprisingly close at 26x cheaper.

[Guillermo Casaus @_guillecasaus]

👋🏼 ADIÓS A CAPCUT.

Ahora puedes convertir un vídeo largo en varios Shorts listos para publicar.

Y sí, lo hace automáticamente.

Se llama OpenShorts y es open source.

¿Lo mejor?

→ Encuentra los mejores momentos del vídeo
→ Genera clips en formato vertical
→ Añade subtítulos automáticamente
→ Prepara contenido para Shorts, TikTok y Reels

Todo dentro de una misma herramienta.

Ideal para reutilizar podcasts, entrevistas o vídeos largos sin tener que revisar y editar cada fragmento desde cero.

Repo en comentarios 👇

[Ozgur Guler @ozguramdin]

Ahenk'i yayınladık: açık ağırlıklı, Türkçe metinden sese (TTS) modeli.

Sayıyı, tarihi, fiyatı ve telefon numarasını doğru okuyor. "mı" ile biten sorularda sesi yükseliyor. Talimatla kızgın, neşeli ya da şaşkın konuşuyor, gülebiliyor.

Videodaki bütün sesler Ahenk'in. Sesi açın.

[Leon Lin @LexnLin]

earshot v0.2 is out now!

made a lot of changes and it can now turn a sound into code.

https://t.co/IQ9TLIvbiu

[Marco Franzon @mfranz_on]

Left: YOLO26n fine-tuned 50 epochs on PPE data

Right: stock YOLO26n finds people, a local 27B vision model (Jev) judges each crop. Zero-shot, no training.

This is probably the future of computer vision.

[Higgsfield AI 🧩 @higgsfield_ai]

DeepSeek V4.1 Flash looks like the new SOTA for motion design.

Look at the 3D detail and UI transitions.

Made with Higgsfiled.

[Tom Huang @tuturetom]

我去，DeepSeek V4.1 Flash 1017 做教程类动效视频已经这么牛了吗，感觉和Opus 5.5没啥区别了😂

我们用它在 OpenDesign 的 Motion Design 里做了一条教程，教大家怎么在其他 coding agent 里接入 OpenDesign API，用上它和其他模型，效果简直不要太好了

[taras @voidisomorphism]

built an arcade to play with the different system one providers out there

attention is all you need? here's how clef, jev and luna decide what they should pay attention too

clef find me boring... luna not

try it out at https://t.co/HSlCxjvTTg more in 🧵

[Greg Pstrucha @grichadev]

so yes, it stand very well against different vision models for people detection.

it's a bit of imperfect science, some of the tools for people detection are meant to be done on video stream, not on video frames, but the fact jev can do it on a single video frame, and with that level of accuracy for both finding people, matching their faces and recognizing known folks (i.e family members) is quite impressive.

[Xunzhuo @XunzhuoLiu]

Turn any camera into a stream of decisions.

Every second, d3-edge (0.6B) watched this rescue and answered 8 plain-English questions in ~0.1 s.

New questions, new scene: pet cam, bird feeder, doorstep. What would you ask?

https://t.co/GBVE6aUdz8

#Jev #Decision

[Maxime Labonne @maximelabonne]

Small decision models can be insanely good when fine-tuned.

We played with d1-omni-600M and it can destroy any model on the games + format (state, input) you train it on.

[Mehmet Eren Dikmen @ErenAILab]

Jev’in yeni güncellemesiyle birlikte artık realtime görüntü işlemek mümkün olabiliyor.

Önceden sadece bu videoda birisi var mı yok mu diye sorabiliyorduk.

[José Siles | AI | Data @josesilesdata]

UN CHINO HA REVENTADO EL OCR:

1,2 MILLONES DE DESCARGAS EN UN MES

Baidu acaba de liberar Unlimited-OCR.

Gratis, pesos abiertos, licencia MIT.

¿Qué hace?

Le metes un documento, una foto, un escaneo, un PDF entero y te lo devuelve como texto limpio y estructurado.

Tablas, fórmulas, el orden de lectura, todo.

Lo diferente está en el nombre: Unlimited.

Los modelos de OCR se ahogan con documentos largos.

Este está hecho para tragarse un PDF entero de una sola pasada, página tras página, sin perder el hilo.

Ellos lo llaman “parsing de horizonte largo”.

→ Documentos con texto, tablas y fórmulas, en estructura
→ Multipágina y PDF de corrido
→ Multilingüe
→ Solo 3.000 millones de parámetros: ligero, corre en una GPU normal
→ Y lo puedes levantar en local con vLLM, SGLang, Ollama o LM Studio

Es la respuesta de Baidu a DeepSeek-OCR, y lo dicen abiertamente: querían llevarlo un paso más allá.

4.400 likes en Hugging Face. 1,2 millones de descargas el mes pasado. MIT.

Link en los comentarios 👇🏻

[0xMarioNawfal @RoundtableSpace]

You can now run the biggest open AI models on a 4 GB graphics card.

AirLLM loads a model one layer at a time instead of all at once, so it fits in a tiny amount of video memory:

→ Kimi K3, 2.8 trillion parameters, in 3.72 GB
→ DeepSeek-V3, 671 billion parameters, in about 12 GB
→ 70B models on a single 4 GB card, without quantizing them
→ It can now fine-tune a 125B model in under 6 GB, too

pip install airllm, then load almost any model from Hugging Face in one line.

It trades speed for memory, and you need plenty of disk space for the model files.

38K+ GitHub stars. Apache-2.0.

https://t.co/jnTebGBnB7

[Muhammad Rizwan Munawar @muhammdrizwanmr]

Bread counting on a conveyor belt using @ultralytics YOLO26 🥪

This system can offer a detailed report on bread quantity, enhancing production and inventory management.

You can also use it for multi-object detection to distinguish among various products, sizes, and packaging types.

#MLops #MachineLearning #Logistics

[Pratik Gandhi @pratikg]

🚨 JUST LAUNCHED: Bonsai 2 from @PrismML.

Open-weight ternary models in 2B, 4B and 9B, small enough to run on a phone. The 2B hits 90 tokens/sec on iPhone 17 Pro Max, with reasoning, vision and tool calling all on-device.

You can try Bonsai 2 9B today on @DarkbloomAI, where the first 100 people get 250M free tokens.

And we’re making it faster on @YukonResearch.

Bonsai 2 9B runs at 170+ tokens/sec at launch.

There are two tracks, Apple Silicon on https://t.co/uFpHiT9o9m and NVIDIA DGX Spark on https://t.co/tuLkHUh7Qx.

To join, pick a track, point your agent at it and submit a faster run.

Every new best is open for anyone to build on.

[Alex Bodner @AlexBodner_]

How do we do to track an object accurately with a camera moving? Meet Camera Motion Compensation!

In practice, you can use BoT-SORT in trackers with a 1 line change:
https://t.co/GPplmZJxmX

[0xSero @0xSero]

Do you have 2 to 4 GB of vram?

Great! Now you can run a lil guy

[Adrien Grondin @adrgrondin]

The new Bonsai 2 2B model reaches up to 120 tk/s on iPhone 18 Pro

It’s better than Qwen 3.5 2B (4-bit) and runs twice as fast while using almost half the memory

[AIGCLINK @aigclink]

OpenAI Realtime API开源替代方案，蚂蚁和清华搞了一个：Realtime-Venus，支持实时音视频感知、全双工语音、异步Agent执行

一套能持续听、看、说，同时在后台执行复杂任务的实时多模态Agent系统

这对智能眼镜、车载助手来说比较有价值，把实时Agent从一个语音接口变成一个能持续观察环境、随时交流、同时处理任务的助手

双循环架构，它把实时交互和耗时任务拆成两条并行运行的链路，核心由两个9B模型和一个Harness组成

Realtime-Venus-Omni，用来持续感知音视频输入
Realtime-Venus-Audio，原生全双工语音
Realtime-Venus-Harness，把复杂推理、搜索和工具调用委派给后台执行，完事后再把结果带回当前对话

前台持续感知和交流，后台异步执行耗时任务，感知上用户不必为了等待AI完成任务中断对话

#RealtimeVenus #OpenAIRealtimeAPI