At least until (if?) rapid improvement stops, it seems less likely someone is going to catch the Big Three AI Labs. Microsoft and Meta released their models, which were fine, but not frontier. SpaceX also hasn't regained its position. Chinese models are improving, but still lag.
AI
-
OpenAI: New Memory Architecture with Automatic “Dreaming” for Context Synthesis
By
–
OpenAI’s new memory architecture moves beyond “saved notes” to an automatic dreaming layer that quietly synthesizes context across many chats. It optimizes for 3 things that actually matter in real work: Carrying forward useful context over time Following your preferences and
-
ChatGPT Dreams: Your AI Teammate Evolves with Your Past Conversations
By
–
ChatGPT doesn’t just “remember” you anymore — it dreams about your past conversations to build a living, evolving profile of your preferences, projects, and constraints. This new dreaming-based memory makes ChatGPT feel less like a chatbot and more like a true AI teammate.
-
Flow-OPD distills specialized teachers into one student model
By
–
What if your text-to-image model could master every task without trade-offs? Researchers from USTC, UCLA, CUHK, and Xiaohongshu present Flow-OPD. They distill specialized teachers into one student via on-policy learning, plus a regularizer to protect image quality. Result:
-
Local models on phone using borrowed MacBook compute for free
By
–
This is the FUTURE.
— Shubham Saboo (@Saboo_Shubham_) 5 juin 2026
Local models running on your phone with borrowed compute from your MacBook.
100% free and works with all Open Source models. pic.twitter.com/bhtzdlbQezThis is the FUTURE. Local models running on your phone with borrowed compute from your MacBook. 100% free and works with all Open Source models.
-
Improving Codex Cli with multi-agent delegation and memory
By
–
Let me make your Codex Cli experience better with – Multi-agent delegation
– Enhanced memory
– Better artifacts
– Children AGENTS.md contextualization
– Runtime metrics (optional) Run the command in the screenshot below to get them up and running Let me know how you like it -
Scobleizer built alignednews.com AI with Supabase and agent
By
–
I built https://
alignednews.com/ai @blevlabs built me a database with Supabase then built me an agent. I took over from there -
AI compute demand strong, NVIDIA platform powers expanding AI clouds worldwide
By
–
AI compute demand isn't slowing down, and neither is the ecosystem to support it. AI Clouds are built on NVIDIA’s full-stack, end-to-end AI factory platform. They’re expanding worldwide to bring accelerated computing closer to developers, enterprises, startups and nations
-
Gary Marcus notes AGI harder than RSI; RSI is useful coding tool
By
–
Critical context on the new Anthropic blog: 1, AGI is *harder* than RSI (as used below). AGI: machine can do anything human can do, autonomously [not achieved] RSI (as used below): AI is a useful coding tool that humans can leverage [achieved]; it’s great at (some) code
-
NEMOTRON 3 ULTRA: Open Source Frontier Intelligence Handbook with 1M Context
By
–
NEMOTRON 3 ULTRA IS HERE
MoE – 550B, 55B Active – Base and Instruct in BF16
– NVFP4 Instruct (350GB)
– 1M Context Window
– Supports MTP Spec Decoding This release IS NOT just about the Weights
It's an Opensource Frontier Intelligence Handbook – Training Data (Pre & Post)
–