Sometimes, a single LLM isn’t enough, and we could coordinate multiple models to solve complex tasks together. Router-R1 is a reinforcement learning–based framework that routes and aggregates multiple LLMs like an intelligent conductor. Key ideas: – Formulates multi-LLM
AGENTS
-
OpenAI Agents Are Fine-Tunes: Operator and Deep Research
By
–
All OpenAI agents are fine tunes: Operator, Agent Mode, Deep Research (same goes for Google), Codex now too. I think even Canvas was too at some point.
-
Claude Code Web Preview: Split Screen Coding Sessions
By
–
BREAKING 🚨: Early Preview of Claude Code for web! Users will be able to choose between different environment types, create and browse coding sessions with Claude.
— 🚨 AI News | TestingCatalog (@testingcatalog) 17 octobre 2025
UI comes with a split screen, and Claude Code functionality will partially replicate what is available on CLI. pic.twitter.com/FDuZDxBzYFBREAKING : Early Preview of Claude Code for web! Users will be able to choose between different environment types, create and browse coding sessions with Claude. UI comes with a split screen, and Claude Code functionality will partially replicate what is available on CLI.
-
Codex Agent Generated OpenAI.fm Designs Applied Locally
By
–
Here’s the exact moment @gpeal8 locally applies the new http://
OpenAI.fm designs that the Codex agent generated in the cloud! -
LangGraph: Building Production-Ready AI Agents with Minimal Abstraction
By
–
In this blog piece, you’ll learn why and how we built LangGraph for production agents. Building upon feedback from the super popular LangChain framework, we aimed to find the right abstraction for AI agents, and decided that was little to no abstraction at all. Instead, we
-
Claude Multi-Agent Systems: Skills, Code, and Future Insights
By
–
Very interesting convo with @ErikSchluntz (multi-agent research lead) about all things agents:
— Alex Albert (@alexalbert__) 17 octobre 2025
– Why Claude is good at agent tasks
– Evolution from workflows to multi-agents
– Tips for creating Skills
– How code unlocks all other domains
– How to use subagents
– The future of… pic.twitter.com/NshRStZVdsVery interesting convo with @ErikSchluntz (multi-agent research lead) about all things agents:
– Why Claude is good at agent tasks
– Evolution from workflows to multi-agents
– Tips for creating Skills
– How code unlocks all other domains
– How to use subagents
– The future of -
GPT-5 Breakthrough: Multi-Model Message Routing for Optimization
By
–
The main breakthrough of GPT-5 was to route your messages between a couple of different models to give you the best, cheapest & fastest answer possible. This is cool but imagine if you could do this not only for a couple of models but hundreds of them, big and small, fast and
-
Stanford Hosts First AI Agents Conference as Primary Authors
By
–
Stanford researchers are organizing the first-of-its-kind conference, where AI agents are the primary authors and reviewers. Join the #agents4science online conference on Oct. 22: https://
agents4science.stanford.edu -
RL-100: Advanced Robotic Manipulation Using Real-World Reinforcement Learning
By
–
RL-100
— AK (@_akhaliq) 17 octobre 2025
Performant Robotic Manipulation with Real-World Reinforcement Learning pic.twitter.com/gETzj0llssRL-100 Performant Robotic Manipulation with Real-World Reinforcement Learning