Google is rolling out Gemini for Chrome to US users! Gemini in Chrome can now understand the context of what you’re doing across multiple tabs, answer questions and integrate with other Google services.

"We’re also developing more advanced agentic capabilities for Gemini in Chrome that can perform multi-step tasks for you from start to finish, like ordering groceries. You’ll remain in control as Chrome handles the tedious work, turning 30-minute chores into 3-click user journeys."

BREAKING 🚨: Notion launched Notion 3.0 with personalised AI agents which can take over all your Notion tasks.

- Deep research & connectors
- AI Agent Memory & personalisation
- And a lot more!

BREAKING 🚨: Notion is launching Custom AI Agents that can operate in the background independently as well as self-update their own memory!

Available under early access, opt-in below 👀

Gemini users can now share custom GEMs with others

Google now lets Gemini users share custom-built Gems—automated workflows and routines—with specific individuals or groups. With Drive-style access controls, this update supports collaborative project management and AI-powered task coordination.

🗞 #gemini

Luma AI released Ray 3, a first reasoning video gen model with studio-grade HDR support and a new draft mode.

Testing time! 👀

Already available 🔥

Cyberpunk robot test 👀

Jules Agent now talks like a Pirate! 🏴‍☠️

OpenAI keeps working on a "Golden Hour" feature and has started gearing up for an upcoming announcement!

The "Manage" button leads to a list of saved Memories.

What would you wish it to do? 👀

Teaser 👀👀👀

Microsoft tests 40 3D Portraits powered by VASA-1 for Copilot

Microsoft is preparing to launch Portraits, an avatar-based voice chat feature within Copilot Labs, targeting select users in the US, UK, and Canada. Backed by VASA-1, it supports real-time animated responses for educational and training purposes.

🗞 #microsoftcopilot

xAI is preparing to release "grok-4-mini-reasoning" and "grok-4-mini-non-reasoning" models on the API.

2M context window 🤯

These models are both available as stealth models on OpenRouter - Sonoma Dusk Alpha and Sonoma Sky Alpha, respectively.

Windsurf released code-supernova model with image input support and a 200k context window!

Available for free for a limited time 👀

* Also available in Cursor

Price & Intelligence comparison 👀