Gemini Robotics 2 is set to introduce whole body intelligence to robots, building upon its foundation of bringing AI agents into the physical world. The upcoming Gemini Robotics ER 2, slated for July 2026, will power robotics with advanced video understanding, sophisticated task orchestration, and multi-robot collaboration capabilities. This iteration follows Gemini Robotics-ER 1.6, expected in April 2026, which focuses on enhanced embodied reasoning for real-world tasks.
Gemini Robotics ER 2 is an advanced "embodied reasoning" model designed to enhance robotic capabilities with video understanding, real-time decision-making, and multi-robot collaboration. It allows robots to plan multi-step tasks, track their own progress through continuous video feeds, adapt to changes, and verify task completion before moving on. This model represents a significant upgrade in orchestrating complex physical AI tasks without pauses, as demonstrated by its use with Boston Dynamics Spot robots. Gemini Robotics ER 2 is now publicly available to developers via the Gemini API, Google AI Studio, and Gemini Enterprise Agent Platform.
Cursor successfully configured a cloud agent environment to enable autonomous testing within their monorepo, treating the development environment itself as a product for agents. This involved making their dev setup agnostic for Linux VMs, enhancing security features for agent environments, and developing a CLI tool called `anydev` to simplify complex build and run commands. These changes, coupled with a self-healing environment system, significantly improved agent efficiency. As a result, cloud agents now author over half of all merged pull requests, demonstrating the value of a robust, agent-friendly development setup.
Google has concluded its fourth six-week Flow Sessions program, an initiative where artists use Google Flow, a creative studio featuring advanced generative models, for co-creation. This program provides creators with the time and space to develop passion projects. The short films produced by the latest cohort of artists are now available for viewing.
Google is launching Lyria 3.5, its newest music generation model, today in Google Flow Music. This update introduces significant advancements designed to help users create richer tracks, offering improved musicality with more complex melodic structures and enhanced lyrics that better adhere to prompts. Additionally, Lyria 3.5 delivers more realistic and emotionally nuanced vocals, along with greater creative control over song tempo and duration.
Google is launching Lyria 3.5, its newest music generation model, within Google Flow Music starting today. This update brings significant advancements across musicality, allowing for richer and more complex melodic structures. Users will also benefit from enhanced lyric generation with improved prompt adherence and more realistic, emotionally nuanced vocals. Additionally, Lyria 3.5 offers greater creative control over tempo and duration, empowering users to craft higher quality, personalized songs.
Claude has announced the release of MCP 2026-07-28, the fifth specification for the Model Context Protocol, which is now being rolled out across Claude products. This significant update introduces a stateless core, moving MCP to a request/response model that simplifies server deployment on serverless and edge infrastructure, improving scalability. It also hardens authorization to align with enterprise OAuth 2.0 and OIDC deployments and graduates official extensions like MCP Apps and Tasks, providing a formal path for developers to add new capabilities. These changes aim to make building and scaling AI agents connected to applications easier and more secure for developers.
Researchers at Anthropic, using Claude Mythos Preview, have discovered improved attack methods against cryptographic algorithms. The AI significantly weakened HAWK, a post-quantum digital signature candidate, effectively halving its key strength after only 60 hours of work. Furthermore, it identified a new, faster attack on a round-reduced version of AES, speeding up previous attacks by 200-800x. While these findings demonstrate AI's potential to uncover mathematical flaws in cryptography, neither result currently affects any production systems as HAWK is not deployed and the AES attack is on a weaker variant.
Google and Japanese telecom leader KDDI have partnered to launch the AI Startup Support Program in Japan. This initiative aims to accelerate innovative AI-native Japanese startups, reflecting their conviction that Japan is home to visionary AI builders. Selected startups will receive joint equity funding from both the Google AI Futures Fund and KDDI, alongside early access to advanced Google AI models like Gemini. Additionally, participants will gain Google Cloud credits, sovereign Gemini and GPU credits via KDDI's data center, and dedicated technical and business support.
Cursor has launched "Cursor Start," a new plan designed specifically for developers in India, priced at ₹649 per month. This plan offers generous access to Cursor's powerful Grok 4.5 and Composer models, along with more agent requests and always-on cloud agents. Addressing local market needs, it allows payments via UPI and provides pricing tailored for the Indian developer community. This initiative responds to the significant growth of Cursor's user base in India, now its third-largest market.
This article details Anthropic's Claude model family, helping users choose the best model for their specific use case. It explains four key classes: Mythos/Fable for frontier capabilities, Opus for reasoning-intensive enterprise tasks, Sonnet for versatile everyday use, and Haiku for low-cost, high-speed applications. The general recommendation is to start with the most intelligent model, like Mythos or Fable, and then test lower-tier options based on task difficulty, latency, and cost-sensitivity to achieve the ideal balance.
Anthropic product designer Nate Parrott utilizes Claude Design, a tool he built, to rapidly explore, iterate on, and share diverse visual ideas, from product prototypes to slide decks and animations. This beta tool originated to help designers quickly generate HTML-based visuals, evolving to incorporate brand guidelines for consistent outputs. Claude Design now functions as a versatile AI collaboration platform for visual communication and ideation, significantly accelerating the design process by producing rich, interactive HTML outputs. While effective for ideation with existing assets, it is not intended for image generation or logo design.
