Gemini 2.5 Pro sets new AI benchmark and launches on AI Studio and Gemini

Google has introduced Gemini 2.5, its most advanced AI model with built-in reasoning capabilities. The Pro Experimental version features a 1 million-token context window and top-tier performance across benchmarks. It's available through Google AI Studio, with wider access planned.

🗞 #aistudio

Anthropic may soon launch Claude 3.7 Sonnet with 500K token context window

Anthropic is reportedly preparing to upgrade Claude 3.7 Sonnet with a 500K token context window, up from 200K. This expansion could support large datasets and complex codebases without relying on retrieval augmentation. Challenges include computational costs and efficient utilization.

🗞 #claude

Mini Apps 1.2 update brings faster load times and new features to World App

World releases Mini Apps 1.2 with improvements in speed, feedback, and customization, while launching a Developer Rewards pilot to allocate $300K monthly in WLD based on usage, aiming to support growth within the World Network ecosystem.

🗞 #worldcoin

ChatGPT Team gains internal knowledge access with Google Drive connector

OpenAI is releasing a beta Google Drive integration for ChatGPT Team, allowing access to real-time context from Docs, Sheets, and Slides while respecting permissions, aimed at reducing workflow interruptions and supporting informed decisions.

🗞 #chatgpt

Workspace folders coming to Grok as xAI builds better tools

xAI is updating Grok with new features including Workspaces for organizing chats with custom instructions and file support, and DeeperSearch for iOS. Voice mode now offers real-time web access using data from X.

🗞 #grok

Claude may get multi-agent Research Mode with memory and task delegation

Anthropic has renamed its research feature from Compass to Research Mode, introducing early support for multi-agent functionality, including tools like web search, memory, and task delegation via sub-agents within a research flow.

🗞 #claude

Early look at NotebookLM’s updated tool for content sourcing

Google's NotebookLM introduces new tools like Discover Sources, Audio Overviews, and Mindmaps to support research tasks. Future plans include MCP and API integration for external data access and expanded audio options for user customization.

🗞 #notebooklm

OpenAI prepares reasoning slider and memory update for ChatGPT users

OpenAI is testing new ChatGPT features including memory for personalized responses, a notifications feed for updates, a Whisper voice input button on web, and a reasoning slider to adjust model depth, with broader rollout expected soon.

🗞 #chatgpt

Amazon unveiled Nova Act: a new AI model for browser automation

Amazon has introduced Nova Act, an AI model that automates browser-based tasks through precise command workflows. Released as a research preview, it targets reliability and developer control in agentic AI, with broad automation potential.

🗞 #ai

Copilot experiments reveal Deep Research, Podcasts and Actions

Microsoft Copilot is testing new features like Avatars, Pages, Deep Research, and potential podcast creation through a limited Labs access. A sidebar layout update and new widgets are also in development, with Pro users seeing early access.

🗞 #microsoftcopilot

Windsurf Wave 6 introduces one-click deploy and enterprise upgrades for developers

Windsurf's Wave 6 update introduces new features such as Turbo Mode, one-click deploy, Jupyter Notebook integration, and improved memory and terminal management, aiming to support productivity and enterprise requirements.

🗞 #ai

Discover sources feature rolls out in Google’s NotebookLM for all users

Google has added a "Discover sources" feature to NotebookLM, allowing users to find and import curated web content using AI. Powered by Gemini, it supports research, planning, and note-taking directly from the web interface.

🗞 #notebooklm

Google plans new Gemini model launch ahead of Cloud Next event

A new Gemini web update hints at a possible model launch, though timing remains uncertain. Features like scheduled prompts, creative partner, and video generation may debut, with speculation surrounding Gemini 2.5 Pro and upcoming experimental tools.

🗞 #gemini

AI Agents and Deep Research among major Copilot upgrades for Microsoft’s 50th anniversary

Microsoft marks its 50th anniversary with a major event on April 4, where it may unveil updates to Copilot, new features like Vision, Pages, and Discover, and possible Surface-exclusive tools like Local Agents and Generative Layout.

🗞 #microsoftcopilot

Llama 4 brings 10M token context and MoE architecture with 3 new models

Meta launched Llama 4 on April 5, 2025, introducing four AI models focused on multimodal tasks and computational efficiency. Scout and Maverick are live, with Behemoth in training. Access is limited by region, language, and regulatory constraints.

🗞 #meta

Microsoft expands Copilot features to rival ChatGPT and Gemini

Microsoft's Copilot update adds capabilities like web browsing, task execution on websites, live video analysis, and memory retention. New tools support research and desktop functions, though concerns over privacy and performance persist.

🗞 #microsoftcopilot

GitHub integration and Veo 2 update expected soon for Gemini users

Google is adding GitHub integration to Gemini, allowing users to connect repositories for code analysis. Only one repo or local folder can be linked at a time. Veo 2, a top-performing text-to-video model, is also being prepared for launch.

🗞 #gemini

Google launches Agent2Agent protocol to connect AI agents across platforms

Google launched Agent2Agent (A2A), an open protocol for AI agents to collaborate across platforms and vendors. Backed by over 50 partners, A2A standardizes agent communication to automate workflows and support enterprise-scale AI adoption.

🗞 #gemini

Upcoming o4-mini and o3 models hint at next phase of OpenAI’s AI lineup

New references to "o4-mini," "o4-mini-high," and "o3" in ChatGPT's code suggest OpenAI is preparing new reasoning-focused models optimized for STEM tasks, offering users varied performance levels based on cost and speed needs.

🗞 #chatgpt

Claude Max subnoscription offers high-usage and priority features for $100–$200 a month

Anthropic has launched a Max plan for Claude, offering $100 and $200 monthly tiers with expanded usage and priority access to new features. Aimed at power users, the plan has drawn mixed feedback over cost, session limits, and usage clarity.

🗞 #claude