Morning edition — Wednesday 11 November 2026

Google made Gemini 4 Carbon, which it calls its strongest model, available on 10 November to developers, enterprises, scientists and consumers. Two weeks earlier Google had released Carbon only to a set of cybersecurity defenders through its Fairwind Program; Gemini 4 Argon remains the default choice for most coding and knowledge work, Google says.

Carbon is on the paid tier of the Gemini API in Google AI Studio and Android Studio and in Gemini for business, at the same introductory price as Argon, $2 per million input tokens and $10 per million output tokens. It is rolling out to Google AI Ultra subscribers in the Gemini app, AI Mode in Search and Google Antigravity, where Google is previewing a /swarm feature that turns one prompt into a team of coordinated agents.

Google says Carbon ranks first on CursorBench 4.0, Perplexity's WANDR and Zapier's AutomationBench at less than half the cost per task of the next best outside model, and second on a new Multimodal Knowledge Agent benchmark built with Artificial Analysis. On safety, Google reports that its safeguards cut the per-attempt breach rate in simulated cyberattack tests to 0.8%, from 3.1% for Argon, and that Carbon passed 92% of a 375-task agentic honesty benchmark, against 70% for Argon.

Apple is working on a Mac app called Harbor that turns a Mac into a private AI inference server for open-weight models, 9to5Mac reported on 10 November, based on material found by leaker Aaron Perris. Apple has not announced it.

According to the description quoted by 9to5Mac, Harbor downloads and runs models locally and can pool the unified memory of several Macs over Thunderbolt 5 to run models too large for one machine or to serve more requests. 9to5Mac says the app is in a private preview under Apple's Claris brand.

At an event in San Francisco on 10 November, Anthropic launched Claude Meetings, which adds Claude to a video call as a visible participant that follows the discussion, drafts plans, answers questions in the chat and pulls figures from connected work apps while the call is still going. It is in beta for Team and Enterprise plans and works with Google Meet at launch, TechCrunch reports.

Anthropic also put Claude Whiteboards and Claude Sheets into beta on all paid plans: a shared board that Claude organises into timelines with owners and dates, and spreadsheets with tabs, formulas and drop-downs that Claude builds and edits in the cells. Claude Dashboards, in beta since October, is now generally available as Claude Data, with certified queries and live sharing still in beta.

The company also introduced interactive connectors that open full versions of partner apps such as Adobe, Atlassian and HubSpot inside Claude. TechCrunch reports that Claude announces itself in meetings and will not join calls with external guests unless an admin allows it.

OpenAI launched Businesses in ChatGPT in beta on 10 November. Through a Business Hub, a company can claim its profile once verified, update its details, upload product feeds with prices and availability, and see aggregated insights such as mentions, views and website clicks, Reuters reports.

A second tool, Business Agents, lets ChatGPT users chat with an agent that answers from the information and guidelines a business sets. OpenAI's page says it is available at first to eligible e-commerce and online service businesses in the US; Shopify merchants can connect their stores directly.

According to Reuters, OpenAI says it does not share users' conversations or Business Agent chats with businesses, and that insights appear only above minimum thresholds. Businesses can sponsor their agent and use it as the destination for ChatGPT ads, with sponsored placements labelled.

New York Governor Kathy Hochul on 10 November launched the statewide programme of the New York Center for AI Responsibility and Research, based at SUNY Binghamton, to independently test frontier AI systems, do foundational safety research and train AI safety experts. Politico first reported the plan at 8:00 New York time that morning, before the previous edition; the official launch followed later in the day.

Greg Morrisett of Cornell Tech and Jeannette Wing of Columbia University will be interim co-directors and bring together more than 80 researchers across the state, using the computing power of Empire AI, the state's $500 million public-private initiative. The centre was first announced in January with a $30 million gift from donors led by Bloomberg LP co-founder Tom Secunda and an initial $25 million from SUNY.

The state's new DIGIT office also opened a portal where large frontier AI developers must register by 1 January 2027 under the RAISE Act. From January, large developers must publish safety frameworks and all frontier developers must report critical safety incidents; the centre will help review the published frameworks.

Mercor, which pays experts to help train and evaluate AI models, announced a $750 million Series D at a $20 billion valuation on 10 November. The round included an employee tender, and investors include Benchmark, Felicis, General Catalyst, Nvidia, Salesforce, Zoom and funds managed by affiliates of Apollo; Mercor did not name a lead.

Reuters reports that the valuation is double the roughly $10 billion Mercor reached when it raised $350 million in October 2025. Mercor says its network has more than five million experts, and that the money will go into evaluations, post-training and environments where agents can practise enterprise workflows. Co-founder Brendan Foody wrote on LinkedIn that the company went from $1 million to $3 billion in run-rate revenue in 27 months.