Anthropic is quietly preparing a more compact attachment menu for Claude

In addition
- The research icon will be updated to an owl
- Opus 4.1 denoscription changed to "Deep brainstorming model. Consumes usage faster."
- Neptune v6 is out

Sus 🦉

xAI is preparing a video extension feature for Grok Imagine on the web. Also, Grok Imagine will get a new selector to choose between video and image generation.

OpenAI is planning to release 2 new fine-tuned open models, "gpt-oss-safeguard-120b” and “gpt-oss-safeguard-20b”. The official blog post was taken down after publishing.

These models will use reasoning to directly interpret a developer-provided policy at inference time.

BREAKING 🚨: Cursor got a major upgrade to version 2.0 with a multi-agent system and its own coding model!

- Cheetah is now Composer-1, and it is 4 times faster than other frontier models.
- The same prompt can run across multiple models
- Built-in browser for testing
- Voice mode! 🔥

I've been testing it for a while, and that's truly a huge upgrade!

Both "gpt-oss-safeguard-120b” and “gpt-oss-safeguard-20b” are already available on HuggingFace.

Internal moderation eval 👀

The Gemini app has 650 million monthly active users, according to Google's recent Q3 earnings report.

$100B in revenue in one quarter is 🤯

OpenAI released Character Cameos in its Sora 2 mobile app. With Character Cameos, you would be able to generate videos of a given character in the same way as with people.

The Sora mobile app is still limited to the US and Canada.

Email Assistant is now available to Perplexity Pro users for 14 days for free.

It can manage your calendar, draft responses and automatically label all the incoming emails.

Google launches Pomelli, experimental Marketing AI Agent

Google is testing Pomelli, an AI tool in Google Labs designed to help small businesses create marketing campaigns by analyzing their website, extracting brand data, and generating customizable campaign assets. It's limited to select regions for now.

🗞 #aitestkitchen

BREAKING 🚨: Windsurf has released a new SWE-1.5 model that delivers "near-SOTA coding performance" at significantly higher speeds (950 tok/s).

Falcon has been revealed 👀

Chat history on NotebookLM is finally here!

NotebookLM got an 8x larger context window, 6x longer conversation memory (conversation history) and better response quality.

Goal setting in chat config is now available to all users.

Google upgrades NotebookLM with Gemini 1m token window

Google's latest NotebookLM update integrates Gemini's 1 million token context window, enabling deeper data retention, nuanced analysis, and extended conversation threads. New role customization and history features support complex research workflows.

🗞 #notebooklm

Cursor debuts Composer and multi-agent suite in Cursor 2.0

Cursor 2.0 introduces Composer, a coding model tailored for rapid code generation and editing, and a multi-agent interface for parallel development. The release targets advanced teams managing complex workflows and code review tasks.

🗞 #cursor

Grammarly rebrands and launches AI-powered Superhuman GO

Superhuman, formerly Grammarly, has rebranded to reflect its expanded AI suite combining Grammarly, Coda, Superhuman Mail, and the new Superhuman Go—an AI agent platform supporting workflows across writing, scheduling, and communication.

🗞 #grammarly