ICYMI: GitHub Copilot exists on mobile right inside the GitHub app. And Microsoft is working on adding Grok Code Fast 1 to the list of available models.
LLMS
-
Mixture of Experts Architecture Powering Frontier AI Models
By
–
Why are the world’s leading models built on mixture of experts?
— NVIDIA AI (@NVIDIAAI) 29 décembre 2025
Ian Buck, Vice President of Hyperscale and HPC at NVIDIA, unpacks the architecture powering today’s frontier AI and how extreme co-design is driving smarter models at lower cost.
🎧 Listen to the NVIDIA AI Podcast… pic.twitter.com/DBMFNDAIQ0Why are the world’s leading models built on mixture of experts? Ian Buck, Vice President of Hyperscale and HPC at NVIDIA, unpacks the architecture powering today’s frontier AI and how extreme co-design is driving smarter models at lower cost. Listen to the NVIDIA AI Podcast
-
Oracle and Claude: Excellence in AI Systems
By
–
Excellent reading thank you. Love oracle and Clawd.
-
Explaining the internal mechanisms of ChatGPT
By
–
Comment fonctionne ChatGPT ? Si vous voulez comprendre en 4 minutes avant de finir 2025, c’est ici, avec des petits schémas que j’ai faits.
— Defend Intelligence (Anis Ayari) (@DFintelligence) 29 décembre 2025
Je vous ai extrait l’explication vulgarisée d’une de mes dernières vidéos YouTube. Plus d’excuses 😄
On peut voir les mécanismes… pic.twitter.com/hMBrBLVyG2Comment fonctionne ChatGPT ? Si vous voulez comprendre en 4 minutes avant de finir 2025, c’est ici, avec des petits schémas que j’ai faits. Je vous ai extrait l’explication vulgarisée d’une de mes dernières vidéos YouTube. Plus d’excuses On peut voir les mécanismes
-
PayPal and NVIDIA debunk large AI agent myths
By
–
This paper from PayPal and NVIDIA quietly kills the myth that agentic AI needs giant models to work. PayPal just published a research paper showing that their biggest performance win did not come from a better prompt, a bigger model, or a clever orchestration trick. It came from
-
MiniMax M2.1 Powers Robot Dog with 230B Parameter Sparse MoE
By
–
MiniMax M2.1 × Vbot is super-powered in action! This AI agent model (230B params, sparse MoE) controls a physical robot dog seamlessly—virtual training to real-world navigation, coding, & multimodal mastery.
— Futurepedia – Learn to Leverage AI (@futurepedia_io) 29 décembre 2025
Speedy, efficient, rivals Claude Sonnet.#MiniMaxM21 #AIRobotics pic.twitter.com/Hji1KAuIgzMiniMax M2.1 × Vbot is super-powered in action! This AI agent model (230B params, sparse MoE) controls a physical robot dog seamlessly—virtual training to real-world navigation, coding, & multimodal mastery. Speedy, efficient, rivals Claude Sonnet. #MiniMaxM21 #AIRobotics
-
Claude Automates Nanochat ML Experiments End-to-End
By
–
Claude has been running my nanochat experiments since morning. It writes implementations, debugs them with toy examples, writes tests and makes them fail/pass, launches training runs, babysits them by tailing logs and pulling stats from wandb, keeps a running markdown file of
-
Build RAG Chatbots with JavaScript and LangChain
By
–
Build Your Own RAG Chatbot with JavaScript Made by the LangChain Community freeCodeCamp just released a course teaching JS devs to build RAG chatbots overcoming LLM knowledge cutoffs. LangChain.js orchestrates DataStax, OpenAI, and Next.js for real-time retrieval. Watch:
-
Unified Gemini Integration for LangChainJS Simplifies Authentication
By
–
Unified Gemini Integration for LangChainJS Made by the LangChain Community Consolidates 6+ packages into @langchain/google. Simplifies authentication for AI Studio and Vertex AI with cross-platform support and LangChainJS 1.0 multimodal for Gemini 3. Alpha in early January