Gemini 3 Pro benchmarks are wild – Humanity’s Last Exam: 37.5%
– ARC-AGI-2: 31.1% True SOTA
@testingcatalog
-
Gemini 3 Pro Benchmarks Break Records
By
–
-
Poe Adds Group Chats for AI-Human Collaboration
By
–
Poe introduced group chats for up to 200 people to enable collaborative work between AI and humans.
— 🚨 AI News | TestingCatalog (@testingcatalog) 17 novembre 2025
Multi-human AI system 👀 https://t.co/VWgFKK2381 pic.twitter.com/0pY5X5fwrEPoe introduced group chats for up to 200 people to enable collaborative work between AI and humans. Multi-human AI system
-
GPT-5.1 tops ARC AGI 2 benchmark
By
–
GPT-5.1 Thinking High from OpenAI claims a top spot on ARC AGI 2 benchmark and dethrones Grok 4. Was GPT-5.1 undehyped?
-
Gemini 3 tooltip added to AI Studio
By
–
BREAKING : A new Gemini 3 tooltip has been added to AI Studio. Preparation is underway. "For Gemini 3, best results at default 1.0. Lower values may impact reasoning." h/t @MarsForTech
-
Google’s Agentspace live feature for Gemini Enterprise
By
–
BREAKING : Google is working on the Agentspace live feature for Gemini Enterprise. This seems to be a multimodal voice experience where users will be able to hold voice conversations with agents. Also, "For You" homepage personalisation and more!
-
Google Adds Agent Import to Gemini Enterprise
By
–
BREAKING 🚨: Google is working on an Agent import feature for Gemini Enterprise, to let users connect external agents into their own multi-agent workflows.
— 🚨 AI News | TestingCatalog (@testingcatalog) 17 novembre 2025
Also, Agent evaluation tool, scheduled execution, new Connectors and more! pic.twitter.com/HcPZ6vZIrBBREAKING : Google is working on an Agent import feature for Gemini Enterprise, to let users connect external agents into their own multi-agent workflows. Also, Agent evaluation tool, scheduled execution, new Connectors and more!
-
Google’s Multi-Agent Idea Generation System
By
–
BREAKING 🚨: Google is working on multi-agent systems to help you refine ideas with tournament-like evaluation. Each run takes around 40 minutes and brings you 100 detailed ideas on a given research topic.
— 🚨 AI News | TestingCatalog (@testingcatalog) 16 novembre 2025
2 new multi-agents are being developed for Gemini Enterprise:
– Idea… pic.twitter.com/q6ZO7VHidoBREAKING : Google is working on multi-agent systems to help you refine ideas with tournament-like evaluation. Each run takes around 40 minutes and brings you 100 detailed ideas on a given research topic. 2 new multi-agents are being developed for Gemini Enterprise: – Idea
-
FlowithOS: Agentic Workspace & Browser in Beta
By
–
ICYMI: Flowith AI released FlowithOS app for macOS and Windows in public beta, where FlowithOS is a mix of an agentic workspace and a web browser.
— 🚨 AI News | TestingCatalog (@testingcatalog) 16 novembre 2025
The user and AI can operate browser tabs and text notes together in a shared workspace. FlowithOS is also an Agentic browser where… https://t.co/J0XqZOqQp1 pic.twitter.com/7hyxOUIxiAICYMI: Flowith AI released FlowithOS app for macOS and Windows in public beta, where FlowithOS is a mix of an agentic workspace and a web browser. The user and AI can operate browser tabs and text notes together in a shared workspace. FlowithOS is also an Agentic browser where
-
Google accelerates feature development in 2025 with agents and tools
By
–
And on top of that, there will be loads of new tools and agents for both individuals and businesses, to bring the value of these models to everyone. Google clearly has accelerated their feature development during the course of 2025. A year of agents
-
Fear of not keeping up with AI news
By
–
I am afraid we won’t manage to keep up with all AI news this time