Microsoft just added xAI's frontier Grok 4 model to Azure AI Foundry, packing a 128K-token context window, first-principles reasoning via "think mode", and live web search. The enterprise deployment integrates Azure AI Content Safety by default, alongside three specialized Fast models tailored for coding, quick tasks, and reasoning.
Microsoft just launched Agent Mode and Office Agent in Microsoft 365 Copilot, bringing agentic workflows to everyday office tasks. Powered by reasoning models from OpenAI and Anthropic, the system can write and troubleshoot Excel formulas, collaborate on Word documents, and generate complete PowerPoint decks directly from chat prompts. It is rolling out to Frontier program subscribers on the web first.
OpenAI just launched Instant Checkout inside ChatGPT, built with Stripe to let people buy items directly in chat without leaving the app. It kicks off with U.S. Etsy sellers before rolling out to over a million Shopify merchants, alongside an open-source Agentic Commerce Protocol. AI is moving fast from product recommendations to actually closing sales.
Anthropic just released Claude Sonnet 4.5, establishing a new high watermark for coding and agent workflows. The model hits 77.2% on SWE-bench Verified and jumps to 61.4% on the OSWorld computer-use benchmark, while keeping API pricing flat at $3/$15 per million tokens. Alongside the release, Anthropic opened up its Claude Agent SDK to let developers build custom autonomous agents.
DeepSeek just released an experimental model called V3.2-exp that uses a sparse attention method to cut long-context API costs in half. The system pairs a lightning indexer with a fine-grained token selector to ease server load over long prompts. It is already open-weight on Hugging Face, opening the door for third-party benchmarking.
OpenAI just launched parental controls for ChatGPT, allowing parents to link accounts with their teens. Adults can turn on Study Mode, set quiet hours, and disable voice mode or image generation, plus get emergency alerts if self-harm signals are detected. It gives families a direct way to keep AI homework sessions and chats age-appropriate.
Apple quietly deployed an internal AI chatbot code-named Veritas to stress-test advanced capabilities for a major Siri revamp. Running on Apple's new Linwood system that blends in-house models with third-party tech, the assistant is learning to search personal files and trigger in-app actions. The public rollout was pushed to as early as March 2026 after early testing proved unreliable.
Meta just launched Vibes, an AI-generated short video feed inside the Meta AI app and on meta.ai. Users can scroll through community clips, generate fresh videos from scratch, or remix existing posts with new music and styles before blasting them to Instagram Reels. It is Meta's clearest attempt yet to make generative video social.
Google just updated Gemini in Sheets to troubleshoot broken formulas and walk you through step-by-step fixes. It suggests alternative functions, explains why specific formulas fail, and breaks down complex options right from the side panel.
Elon Musk’s xAI just inked a deal with the U.S. General Services Administration to supply its Grok chatbot to federal agencies for just 42 cents over 18 months. That undercuts OpenAI and Anthropic’s $1-a-year bids and even throws in hands-on engineer support. Why it matters: Frontrunner labs are basically giving away frontier tech for pennies just to lock in lucrative long-term government footholds.
OpenAI just rolled out shared projects for ChatGPT business plans, letting teams collaborate with shared instructions, files, and private project memory. The update also adds smarter connectors for tools like Gmail and Teams that trigger automatically, alongside new ISO certifications and IP allowlisting for enterprise security.
OpenAI just launched GDPval, a new benchmark designed to measure model performance on real-world knowledge work across 44 occupations contributing heavily to U.S. GDP. Deliverables span legal briefs, blueprints, and spreadsheets, with frontier models approaching expert quality up to 100x faster and cheaper. Why it matters: It shifts AI evaluation away from synthetic tests and into tangible workplace productivity.
OpenAI just launched a preview of ChatGPT Pulse for Pro subscribers on mobile. Instead of waiting for prompts, it works overnight using your chat history, memory, and connected apps like Google Calendar to serve up daily research cards each morning. It marks OpenAI's shift from reactive chatbots to proactive personal agents.
Microsoft is integrating Anthropic’s Claude models into Office 365 apps like Excel and PowerPoint, breaking its exclusive reliance on OpenAI. Internal tests showed Claude outperforms OpenAI at building slide decks and automating financial tasks. Why it matters: Microsoft is actually paying direct cloud rival AWS to access the models, proving that raw performance now beats exclusive partnerships.
Google just rolled out Search Live to all U.S. users in English without requiring a Labs opt-in. The feature combines real-time voice chat with your phone's camera feed, letting Search see what you're pointing at and talk through troubleshooting cables or recipes. It fundamentally shifts search from text queries to live multi-modal assistance.
Google just expanded its budget-friendly Google AI Plus subscription to 40 additional countries following an initial launch in Indonesia. The tier includes higher limits for image editing, access to the Veo 3 Fast video model across Gemini, Whisk, and Flow, 200 GB of storage, and Gemini inside Gmail, Docs, and Sheets. Why it matters: Google is making its creative and productivity tools accessible at a lower price point across dozens of developing markets.
Google just expanded AI Mode in Search globally to Spanish-speaking users. The update lets millions more people ask complex questions in Spanish and get direct answers backed by web links to dig deeper. It is another big push to take Google's most powerful AI search capabilities well beyond English.
Meta just rolled out an AI assistant for Facebook Dating to save singles from endless profile swiping. Users in the U.S. and Canada can simply prompt the bot with specific preferences—like finding a marketer in New York—while a new "Meet Cute" tool delivers algorithmic surprise matches beyond their usual type.
Alibaba just introduced Qwen3-Omni, a natively end-to-end multimodal model unifying text, images, audio, and video under a single architecture. It boasts a 211ms latency alongside 30-minute audio comprehension, releasing 30B-scale instruct, thinking, and captioner checkpoints for developers.
Perplexity just launched Email Assistant, an AI tool exclusively for its $200-a-month Max subscribers to schedule meetings, prioritize inboxes, and draft replies across Gmail and Outlook. Users simply email a dedicated assistant address to assign tasks, and Perplexity claims the AI mimics personal writing styles without training on private email data.
