跪拜 Guibai
← Back to the summary

Huang Renxun's Nvidia Return Chatter Moves Markets; Huawei Ships an Embodied World Model and OpenAI Wires Up Agent Payments

AI Hot News Daily (2026-11-16): Huang Renxun May Return to Nvidia, Huawei Releases Embodied World Model, ChatGPT Wallet Ties Up Payments

Today's AI core focus: Nvidia's dark-pool stock price surged over 4% in a straight line. The British Financial Times reported that Jensen Huang is considering returning to Nvidia as CEO; Musk and other Silicon Valley allies are lobbying the board to 'bring Huang back'. Domestically, Huawei Cloud released its first Ascend-native embodied world model, CloudRobo WM1.2; Alibaba plans to spin off T-Head for a standalone Hong Kong listing. On the model side, Google launched a new high-speed agentic model, Gemini 4 Hydrogen; DeepSeek opened full-stack app capability APIs. On the application side, OpenAI launched ChatGPT Wallet to enable agent payments, Meta's Muse can make phone calls on behalf of users, and Anthropic's IPO is imminent.

Models and Technology

1. Huawei Cloud Releases First Ascend-Native Embodied World Model CloudRobo WM1.2 🔥

Huawei Cloud released the first Ascend-native embodied world model, CloudRobo WM1.2, today at 14:00. On the same day, the China Academy of Information and Communications Technology (CAICT) focused on embodied intelligence, marking a breakthrough by domestic computing infrastructure in the embodied world model track.

Source: CSDN/Knowledge Planet

2. Google Releases Gemini 4 Hydrogen: 2.5x Faster, Integrated into Search 🔥

Google released a new high-speed agentic model, Gemini 4 Hydrogen, adapted from Gemini 4 Carbon. It completes tasks 2.5x faster than Gemini 3.8 Flash, natively supports a 1 million token context, defaults to a high reasoning level and supports a new ultra-high level. It is now globally available and integrated into Google Search. Pricing is $1.5 per million input tokens and $7.5 per million output tokens.

Source: CSDN/NetEase

3. DeepSeek Opens Full-Stack App Capability APIs, Cached Input Drops to as Low as $0.003/Million Tokens

DeepSeek has API-ized capabilities including web search, file reading, link reading, OCR, and voice. Developers can reuse them through three endpoints: Messages, Responses, and ChatCompletions. It also released an open-source chat demo. V4.1 Flash pricing hits a new low, with cached input as low as $0.003 per million tokens.

Source: CSDN

4. Tencent Hunyuan Releases Real-Time Voice Model Hy Realtime: Supports Tool Calls and Singing

Tencent Hunyuan released a new real-time voice model, Hy Realtime, adding Agentic tool calling and singing capabilities. It can autonomously decide whether to search the web based on context. It is already live in Yuanbao voice calls and Tencent customer service.

Source: Unknown source

Products and Applications

5. OpenAI Launches ChatGPT Wallet to Enable Agent Payments, Dots Agent Can Check Out 🔥

OpenAI launched ChatGPT Wallet, using tokenized card numbers and biometric authorization. The Dots agent can handle shopping checkouts, embedding payment capability into the conversational agent execution chain. Users do not need to expose card numbers to merchants, closing a key link in the agent commerce loop.

Source: Unknown source

6. Meta's Muse Launches Phone-Calling Feature; Sierra and Meta Release Poppy Agent Protocol

Meta's Muse launched a phone-calling feature that can contact merchants or insurance companies on behalf of users. Sierra and Meta released the Poppy protocol to standardize commercial collaboration between personal agents, with agent commercialization continuing to land.

Source: Unknown source

7. Cloudflare Open-Sources AI Agent Browser Kitesurf: Lower Resource Usage Than Chromium

Cloudflare open-sourced Kitesurf, a browser built specifically for AI agents, under the Apache 2.0 license. It can be deployed on Cloudflare Workers. In tests, its CPU and memory usage during screenshots and web page extraction were lower than Chromium.

Source: Unknown source

8. Leju Intelligence Advances Humanoid Robot Commercialization: 15 Units Running Continuously for Over 2 Months Without Failure

On an automotive parts production line built with a listed company, 15 wheeled humanoid robots have been picking and feeding materials continuously without failure for over 2 months, achieving commercial deployment of an embodied intelligence plus automotive parts feeding solution.

Source: Unknown source

Capital and Markets

9. Nvidia Dark Pool Surges 4% as Huang Considers Returning as CEO 🔥

In the early hours of November 16 Beijing time, Nvidia's dark-pool stock price rose in a straight line, with gains exceeding 4%. According to the British Financial Times, Jensen Huang is discussing the possibility of returning to Nvidia as CEO. Executives including Musk and Epic Games CEO Tim Sweeney are asking the board to 'bring Huang back'. This drove U.S. stock futures and other AI hardware stocks higher.

Source: NetEase/Financial Times

10. Alibaba Plans to Spin Off T-Head for Hong Kong Listing; Zhenwu Series AI Chips Have Shipped Millions of Units 🔥

Alibaba announced plans to spin off its wholly-owned subsidiary T-Head for a standalone Hong Kong listing. T-Head's Zhenwu series AI chips have cumulative shipments in the millions. The spin-off marks accelerated capitalization for a leading domestic AI chip player.

Source: NetEase/Sina

11. Anthropic IPO Imminent: Releases Claude Fable 5.2/Mythos 5.2, and Launches Claude for Manufacturing

With Anthropic's IPO imminent, new flagship models Claude Fable 5.2 and Mythos 5.2 top multiple benchmarks with optimizations in performance and cost. Claude Opus 5.5 becomes the default model in paid plans. Also launched is Claude for Manufacturing, which includes 17 manufacturing skills and 4 connectors; the enterprise version is free.

Source: Unknown source

Policy and Ecosystem

12. Shanghai Issues Scientific Data Management Measures: Build AI Data Governance and Product Development Mechanisms

The Shanghai Scientific Data Management Measures (Trial) were issued, clarifying a mechanism for integrating scientific data with AI development, establishing an AI-oriented data governance and product development direction, including high-quality scientific datasets in research evaluation systems, encouraging open sharing, and publishing a unified citywide scientific data catalog.

Source: Unknown source

13. OpenAI, Google, Microsoft, Anthropic Pledge to Support AI Training Opt-Out Token

Four major AI companies have committed to supporting an 'AI training opt-out token'. Websites can declare whether they refuse AI training in one go via a standard protocol, without needing to set rules for each AI. Anthropic has already deployed it, OpenAI plans to implement in the coming months, and Google and Microsoft will integrate by February 2027.

Source: Unknown source

14. U.S.-China Central Bank Hearings Focus on AI: Fed's Productivity and Employment Task Force Holds Live Hearing on AI

The Federal Reserve's Productivity and Employment Task Force held a live hearing on artificial intelligence on November 16 in the San Francisco Bay Area, focusing on AI's impact on productivity and employment patterns, highlighting global regulators' continued attention to AI's economic effects.

Source: Unknown source


This article was automatically generated by AI Hot News Daily. News is sourced from public internet information, and copyright belongs to the original media.