Side quest: Diffusion model but instead image only input/output but for language modeling.
LLMS
-
Comet Browser amélioré avec capacités agentiques
By
–
BREAKING 🚨: Comet browser got a major upgrade for its agentic capabilities.
— 🚨 AI News | TestingCatalog (@testingcatalog) 6 novembre 2025
– It can work longer and on more complex jobs across larger periods of time.
– It performs 23% better than its predecessor.
Longer context was a top request from all the users! Now it can write a… https://t.co/VsTPIoiok4 pic.twitter.com/bJPazyk69fBREAKING : Comet browser got a major upgrade for its agentic capabilities. – It can work longer and on more complex jobs across larger periods of time.
– It performs 23% better than its predecessor. Longer context was a top request from all the users! Now it can write a -
Kimi K2 Thinking: Breakthrough Open Source Model with 1T Parameters
By
–
Las cifras reportadas por el nuevo Kimi K2 Thinking son un barbaridad para un modelo open source! Además cuantizado a INT4 para mayor velocidad de inferencia y arquitectura tipo MoE. Eso sí, 1T de parámetros con 32B activos, no apto para mucho de nuestros equipos
-
Building a Typescript Deep Research Agent with DeepAgents
By
–
Building a Typescript deep research agent In this video, we will walk through how to easily build a Typescript deep research agent This builds upon our new DeepAgents library All it involves is some detailed prompting, some search tools, and some specialized sub agents
-
Building Streaming Agents in Next.js with LangChain
By
–
Excellent walkthrough of how to build a streaming agent in @nextjs using LangChain. Clear explanation of server-sent events, UI streaming, and memory via thread IDs. If you’re evaluating production agent architectures, this is a strong reference implementation. Watch the
-
Steal Gemini Prompt to Multiply Content into High-Performing Posts
By
–
Steal my Gemini prompt to multiply one piece of content into high-performing posts for any platform. —————————————
CONTENT MULTIPLICATION ENGINE
————————————— You are a content strategist who transforms single pieces of -
MIT Study Charts Legible Path for LLM-Assisted Programming
By
–
While programming w/LLMs seems promising, code can be messy, opaque, & hard to change safely. An MIT case study charts a more legible, modular path forward for software, bringing together features that'd otherwise be scattered across multiple services: https://
bit.ly/4pg2sBJ -
Kimi K2 Thinking Model Released with SOTA Performance on HLE
By
–
BREAKING : Kimi K2 Thinking comes as a SOTA on HLE with 44.9% acheivement and now available on Kimi Chat and APIs. K2 Thinking with full agentic mode coming soon. Testing time
-
DeepSeek-OCR: Revolutionary VLM Reduces Token Usage in OCR Processing
By
–
📢Revolutionizing OCR: How DeepSeek-OCR Solves Token Explosion
— Satya Mallick (@LearnOpenCV) 6 novembre 2025
A picture is worth a thousand words, but processing those words shouldn't cost a fortune in compute!
Enter DeepSeek OCR: A game-changing VLM that compresses complex document visuals into just 64 – 400 tokens.
What’s… pic.twitter.com/DjjGcVgOgBRevolutionizing OCR: How DeepSeek-OCR Solves Token Explosion A picture is worth a thousand words, but processing those words shouldn't cost a fortune in compute!
Enter DeepSeek OCR: A game-changing VLM that compresses complex document visuals into just 64 – 400 tokens. What’s -
Jamba Reasoning 3B: Lightest Model Running on Just 2.25 GiB RAM
By
–
How much RAM do you need to run tiny models? Jamba Reasoning 3B runs on just 2.25 GiB, the lightest among small models like Qwen (
@Alibaba_Cloud
), Llama (
@Meta
), Granite (
@IBM
), and Gemma (
@GoogleDeepMind
). Try Jamba Reasoning 3B yourself: https://
huggingface.co/collections/ai
21labs/jamba-reasoning-3b
…