Anthropic has made Claude Opus 5.5 the default model for all paid plans and released a new guide advising users to abandon legacy prompts like "think hard," recommending instead to start interactions at medium effort levels for improved efficiency.
NVIDIA introduces SOMA Retargeter, leveraging human motion data as a teaching foundation by capturing coordination and timing information to address humanoid robot movement control challenges. The tool aims to simplify the mapping process from human demonstration to robot execution, enhancing development efficiency.
Fei-Fei Li’s World Labs has publicly opened access to its next-generation world model, Atlas. Users can now generate 3D Gaussian splats from images via the Viewfinder web app or create explicit 3D geometry using agentic workflows in Chisel. The company also released all underlying API endpoints, enabling developers to build custom tools or allow agents to interact directly with the models, with outputs including keyframes, video clips, and 3D Gaussian splats.
xAI announces that Grok Bot on Mac now supports seven system apps: Messages, Mail, Notes, Reminders, Calendar, Location, and Contacts. Users can complete cross-app tasks via natural language prompts, such as finding an address and scheduling events. The feature is included in X Premium and SuperGrok subscriptions, requiring manual permission grants and user confirmation for sensitive actions.
Anthropic's official developer account released a video detailing the API launch process for new models and their accompanying features. The core advice highlights that when upgrading to new models like Claude Fable 5.2, developers should not rely solely on official benchmarks. Instead, they must rethink their agent harnesses and measure effectiveness via "cost per task" rather than "cost per token," while comparing correction needs and completion times in real-world business scenarios.
Chrome's developer team has released Modern Web Guidance 1.0, a toolkit that injects 200+ expert-written guides and Baseline browser-support data into AI coding agents like Cursor, Copilot, and Antigravity, steering them to write modern, standards-aligned HTML, CSS, and JavaScript by default. The team's own evals found that even the latest frontier models fail more than 40% of edge cases on modern web best practices.
On 11 November 2026, Google DeepMind introduced the Gemini Decisions API, powered by a new model, Gemini Decision 1 (gemini-decision-1). It is built for bounded judgments such as yes/no, multiple choice, and rubric scoring. One request can ask up to 1,000 independent questions about a shared multimodal input and return probabilities and confidence scores. The announced price is $0.10 per million input tokens, with output tokens billed at zero.
On 11 November 2026, Google said the Gemini app is coming to Apple CarPlay, so drivers can use Gemini Live for hands-free conversations in the car. Officially, it is rolling out globally to all Gemini users on iOS, across every supported Gemini Live language. To start, you need the latest Gemini app on iPhone, microphone permission enabled, and at least one prior Gemini Live session on the phone.
xAI official documentation, Get started, walks through the full path: install the desktop app from x.ai/bot, sign in with a supported plan, hand the Bot a real first task, and schedule it as a routine once the result looks right. This article follows those official steps and notes how approvals, the Vault, and mobile fit in.
The official Grok Bot account announced Grok Bot 1.0, positioned as AI teammates that work alongside an entire team. The product page says it is out of beta, with support for sharing bots, collaborating in Slack and Microsoft Teams, inviting bots to meetings, and bots that learn and become more helpful. Musk quoted the announcement the same day, suggesting hiring Grok @Bot as a colleague.
On 10 November 2026, OpenAI announced a beta for Businesses in ChatGPT. Eligible businesses can claim a profile, maintain product pricing and availability, and see aggregated mention and website-click insights, which OpenAI says do not expose user conversations. Business Agents initially serve eligible US ecommerce and online-service businesses, and business verification does not by itself guarantee permission to publish an agent.