OpenAI Rolls Out ChatGPT Images 2.0,

OpenAI launched ChatGPT Images 2.0, its most capable image generation model yet, designed to handle complex visual tasks and produce precise, production-ready visuals with significantly sharper editing and richer layouts.

personal context store that dramatically improves its ability to assist with coding, debugging, and automation tasks.

OpenAI Launches Chronicle, Letting Codex Build Persistent Memory From Your Mac Screen

OpenAI has introduced Chronicle, a new research preview feature that allows its Codex coding agent to continuously learn from a user’s screen activity, creating a rich,

Moonshot AI’s Kimi K2.6 Tops Open-Source Coding

Moonshot AI, the Chinese startup backed by Alibaba, has released Kimi K2.6, a 1-trillion-parameter open-weight model that is now leading open-source coding performance and narrowing the gap with the best closed frontier models.

Google Opens Google AI Studio to AI Pro and Ultra

Google expanded access to Google AI Studio by allowing subscribers to its consumer AI plans AI Pro at $20 per month and AI Ultra at $250 per month to sign in directly with their Google accounts, eliminating the need for separate API keys.

Grok Imagine Adds Custom Template Creation and Sharing, Turning Users into AI Creative Studio Owners

xAI has rolled out a major new feature for Grok Imagine that lets users create, save, and share custom AI templates with their followers, significantly expanding the platform’s creative flexibility.

Anthropic Releases Claude Opus 4.7,

Anthropic launched Claude Opus 4.7, the latest flagship version of its most powerful AI model, engineered specifically for demanding, multi-step workflows such as advanced coding, agentic tool use, and extended reasoning.

Anthropic Launches Claude Design,

Anthropic introduced Claude Design, a new visual creation capability within Claude that allows users to generate high-quality prototypes, presentation slides, and one-pagers simply by describing what they want in plain language. Powered by Claude Opus 4.7,

app themes for developers, moves designed to make the company's assistant more sticky, more personal, and easier to customize as competition with Microsoft Copilot and Anthropic's Claude intensifies.

Google Deepens Gemini Integration With 'Skills,' Personal Intelligence, and Developer Themes

Google announced a trio of updates to its Gemini AI ecosystem, introducing "Skills" for Chrome, expanding Personal Intelligence globally, and launching AI-generated

Google DeepMind Releases Gemini Robotics-ER 1.6,

Google DeepMind on Wednesday launched Gemini Robotics-ER 1.6, its latest state-of-the-art model specifically engineered to give robots significantly stronger visual, spatial, and task-level reasoning abilities for complex physical environments.

Lovable Payments Launches AI-Powered “Describe-to-Sell” Checkout Platform

Lovable, the AI application builder, has introduced Lovable Payments, a new product that lets merchants create and launch secure payment flows using nothing more than a natural language description of what they want to sell.

marking the most aggressive challenge yet to Microsoft's own Copilot in the battle for enterprise AI productivity.

Anthropic Takes Aim at Microsoft Copilot with Native Claude Integration for Word

Anthropic launched Claude for Word, a native sidebar integration that embeds the company's AI assistant directly into Microsoft's flagship document editor,

marking the company's push into the rapidly commoditizing layer between foundation models and end-user applications.

Anthropic Enters Agent Infrastructure Race With 'Claude Managed Agents' Beta

Anthropic today launched Claude Managed Agents, a new service that provides businesses with the infrastructure needed to build and deploy AI agents at scale,

World Labs Upgrades 3D AI With Marble 1.1

World Labs, the spatial AI startup, rolled out Marble 1.1 and Marble 1.1-Plus, incremental but significant upgrades to its 3D world-generation model that improve visual fidelity and enable larger, more complex virtual environments.

as competition intensifies with OpenAI, Google, and Anthropic. The model, available immediately at meta.ai and in the Meta AI app, supports tool-use, visual chain-of-thought reasoning, and multi-agent orchestration capabilities.

Meta Enters Multimodal Reasoning Race With 'Muse Spark,' First Model From Superintelligence Labs

Meta Platforms Inc. unveiled Muse Spark, the first release from its newly formed Meta Superintelligence Labs, marking the company's most ambitious push yet into natively multimodal AI reasoning

Pika Labs Releases PikaStream 1.0,

Pika Labs launched the beta of PikaStream 1.0, a new real-time video generation model that powers the first universal video chat skill for AI agents, allowing them to join conversations with a face, voice, memory, and personality.

Google DeepMind Launches Gemma 4 Open Models

Google DeepMind introduced Gemma 4, its most capable family of open models to date, purpose-built for advanced reasoning, agentic workflows, and on-device deployment across a wide spectrum of hardware from smartphones to workstations.

Z.ai Launches GLM-5V-Turbo, High-Performance Vision-to-Code Model

Z.ai introduced GLM-5V-Turbo, a new multimodal model optimized for converting images, videos, design drafts, and screenshots directly into executable, production-ready code.

with a new NO_FLICKER mode designed to eliminate long-standing screen flickering during real-time interactions.

Anthropic Introduces NO_FLICKER Mode for Claude Code Terminal Agent
Anthropic has released version 2.1.89 of Claude Code, its open-source agentic coding tool that runs directly in the terminal,

Google Launches Affordable Veo 3.1 Lite Video Generation Model

Google on Tuesday introduced Veo 3.1 Lite, a new cost-optimized addition to its Veo 3.1 family of AI video generation models, aimed at developers and businesses seeking high-volume production without premium pricing.

Alibaba Unveils Qwen3.5-Omni,

Alibaba’s Qwen team on Monday released Qwen3.5-Omni, the latest generation of its fully omnimodal large language models that natively process and reason across text, images, audio, and video in a unified architecture.

designed to deliver higher-accuracy, better-sourced reports by orchestrating multiple frontier AI models in a generation-plus-evaluation workflow.

Microsoft Introduces Critique, Multi-Model Deep Research System in M365 Copilot Researcher

Microsoft Corp. launched Critique, a new multi-model deep research capability inside Microsoft 365 Copilot’s Researcher agent,

Pretext Library Cracks Web Text Layout Bottlenecks With Canvas-Powered Math

A sleek new open-source library called Pretext is gaining rapid traction among frontend developers by eliminating one of the web’s longest-standing performance headaches: dynamic multiline text measurement and layout.