Still rolling out, so some users may not see it right away. This is very much a prototype to help us learn more about how people feel talking to an AI – if you try it out, please send us your feedback!
INNOVATION
-

Copilot Portrait: Real-Time Animated AI Face Conversations
By
–
You can now talk to a Copilot Portrait in real-time! We’ve heard from some users that they'd feel more comfortable talking to a face when using voice. So in the US, UK + Canada, we’re rolling out a new Copilot Labs experiment where you can talk with animated portraits.
-
AI Agents Now Capable of Real Valuable Work
By
–
AI agents are now capable of doing real, if bounded, work. But that work can be very valuable. For example, the new Claude Sonnet 4.5 was able to replicate published economics research from data files & the paper. We need to figure out what to do with it:
-
Quantum Computing’s Paradox: Building Systems Easier Than Simulating
By
–
Quantum computing’s killer app is supposedly simulating quantum systems, but it’s easier to build the real system than the quantum computer to simulate it.
-

Anthropic Releases Claude Sonnet 4.5, World’s Best Coding Model
By
–
BREAKING: Anthropic just released Claude Sonnet 4.5, calling it the world’s best coding model • Achieved 82% accuracy on SWE-bench Verified, topping GPT-5 and Gemini 2.5 Pro • Strongest yet at building complex AI agents and using real computers for tasks • Claude Code
-
AI Models Execute Bash Tasks Through Agent-Based Paradigm
By
–
coding generalizes well give a model python and let it execute tasks in bash basically they're focusing on agents to lead to AGI, and betting on a very specific paradigm that's all
-

AI providers become all-in-one, sites deserted
By
–
And there it is, the great march of AI providers to become all-in-one apps has begun. Eventually, websites will be deserted: everything will go through AI. If you want to know how much your products are recommended by AI models, now is the time to get started.
-

Claude Code 2.0 Checkpoints: New AI Model Features
By
–
En cualquier caso, como siempre, a estos modelos hay que saborearlos y la mejor forma es dentro del nuevo Claude Code 2.0 (que de novedad ahora incluye checkpoints !!) y ver si con eso vuelven a convencer a parte de la comunidad que había migrado a Codex en las últimas semanas.
-

Claude Sonnet achieves 77.2% on SWE-bench Verified
By
–
En capacidades agénticas, que es donde estos modelos tienen un mayor impacto cuando se orquestan con el uso de herramientas o integrándose en Claude Code, Sonnet logra ahora un nuevo record em SWE-bench Verified, con un 77,2% Parece poco, pero estas pequeñas ganancias importan!
-

LongLive: Real-time Interactive Long Video Generation Technology
By
–
LongLive
— AK (@_akhaliq) 29 septembre 2025
Real-time Interactive Long Video Generation pic.twitter.com/oSPtiqZoGWLongLive Real-time Interactive Long Generation