Guozhen AIGlobal AI field notes and model intelligence

Realtime AI News

Track model releases, AI agents, open-source projects, infrastructure, policy, and product updates as they publish.

Read details: OpenAI embeds Apollo Research evaluators to probe whether its models try to subvert oversight
OpenAI / Apollo Research / AI Safety

OpenAI embeds Apollo Research evaluators to probe whether its models try to subvert oversight

OpenAI will station six Apollo Research safety evaluators inside the company as a pilot, investigating whether its models attempt to undermine the alignment and control mechanisms that keep them, or their successors, in check. OpenAI is granting scoped access to training code, datasets, reinforcement-learning rollouts and intermediate checkpoints while reserving the right to withhold material on security or IP grounds, leaving the pilot's independence in question.

Read details: OpenAI Expands Free Frontier-Model Access to 200,000 Academic Researchers
OpenAI / Research / Frontier Models

OpenAI Expands Free Frontier-Model Access to 200,000 Academic Researchers

OpenAI is expanding its ChatGPT for Academic Researchers program to give 200,000 researchers worldwide free access to its frontier models. The announcement emphasizes reach rather than new capabilities, and leaves the eligibility rules and covered model versions undisclosed.

Read details: Qwen4-Max goes official with 2.4T MoE parameters and a claimed 60% cost cut
Alibaba / Qwen / Model Release

Qwen4-Max goes official with 2.4T MoE parameters and a claimed 60% cost cut

The official version of Alibaba's Qwen4-Max is live, with Sohu reporting a 2.4-trillion-parameter mixture-of-experts architecture and task costs roughly 60% lower. Related coverage also cites a 1-million-token context window for long documents and complex workloads.

通义千问Qwen4-Max正式版上线:2.4万亿参数MoE,任务成本降近六成
Read details: Report: OpenAI Explored Buying Interpretability Startup Goodfire for Over $5 Billion
OpenAI / Acquisition / Interpretability

Report: OpenAI Explored Buying Interpretability Startup Goodfire for Over $5 Billion

OpenAI explored acquiring the AI interpretability startup Goodfire in a transaction valued at more than $5 billion, The Information reported. The figure, if accurate, would rank among the largest sums ever attached to a company working on AI interpretability.

Read details: Lambda Adopts NVIDIA Photonics CPO Switches to Harden 10,000-GPU AI Factories
NVIDIA / Lambda / AI Infrastructure

Lambda Adopts NVIDIA Photonics CPO Switches to Harden 10,000-GPU AI Factories

Lambda is deploying NVIDIA Photonics co-packaged optics (CPO) switches to connect NVIDIA GB300 NVL72 racks carrying more than 10,000 GPUs. The cloud provider is betting that a faster, more efficient and more resilient network is what keeps very large training clusters productive.

Lambda 采用 NVIDIA 光电共封装交换机,为万卡 AI 工厂加固网络
Read details: NVIDIA Extends Slurm to Put Quantum Processors Inside Supercomputing Workflows
NVIDIA / Quantum Computing / Supercomputing

NVIDIA Extends Slurm to Put Quantum Processors Inside Supercomputing Workflows

NVIDIA is expanding Slurm's support for quantum processors so supercomputing centres can schedule quantum hardware alongside their existing scientific workloads. The move lands as nearly 30 centres prepare to deploy more than 50 quantum processors, a roughly sevenfold rise in deployments over five years.

NVIDIA 扩展 Slurm 支持,把量子处理器接入超算调度体系
Read details: Nvidia's Vera Rubin debuts in MLPerf with a training sweep and 10x agentic efficiency gain
NVIDIA / MLPerf / Infrastructure

Nvidia's Vera Rubin debuts in MLPerf with a training sweep and 10x agentic efficiency gain

Nvidia published Vera Rubin's first MLPerf results, claiming leadership across training, post-training and agentic inference, and a sweep of every MLPerf Training v6.1 benchmark. In its first MLPerf Endpoints submission, the platform reports more than 10x generational gains in throughput per watt for agentic AI.

英伟达Vera Rubin首秀MLPerf:训练基准全数拿下,智能体推理能效提升逾10倍
Read details: Nvidia expands CUDA-X with cuPhoton, cuPotential and cuSpa for AI research agents
NVIDIA / CUDA-X / Scientific Computing

Nvidia expands CUDA-X with cuPhoton, cuPotential and cuSpa for AI research agents

Nvidia expanded its CUDA-X libraries with cuPhoton, cuPotential and cuSpa, aimed at research in astrophysics, materials science and single-cell biology. The company says the speedups let scientists analyse larger datasets and model more complex systems, while giving AI agents callable tools for discovery.

英伟达扩充CUDA-X库:cuPhoton、cuPotential、cuSpa把科学计算交给AI智能体
Read details: CUDA at 20: Nvidia marks two decades of GPU computing at SC26
NVIDIA / CUDA / GPU

CUDA at 20: Nvidia marks two decades of GPU computing at SC26

Nvidia is marking the twentieth anniversary of CUDA this week at SC26, honoring the researchers, developers and engineers who built on its GPU programming model, according to a company blog post published November 16. The retrospective frames two decades of CUDA as the story of how a graphics platform became the default substrate for scientific computing and AI.

CUDA二十周年:英伟达在SC26回望GPU计算如何成为科学与AI的地基
Read details: Nvidia unveils Rubin HPC, a 50,000-core-plus GPU built for science agents
NVIDIA / GPU / HPC

Nvidia unveils Rubin HPC, a 50,000-core-plus GPU built for science agents

At SC26, Nvidia announced Rubin HPC, a new GPU in its Vera Rubin platform aimed at simulation and data processing for science, engineering and analytics, with more than 50,000 CUDA cores on a single chip. Nvidia says the card scales data movement alongside compute, paired with sparse-data acceleration and seventh-generation NVLink, to shorten the loop from an agent's hypothesis to a simulated result.

英伟达发布 Rubin HPC:单卡逾 5 万 CUDA 核心,为科研智能体加速仿真
Read details: Audible turns to AI audio previews to help listeners pick audiobooks
Audible / AI audio / Content discovery

Audible turns to AI audio previews to help listeners pick audiobooks

Audible has added AI-generated audio previews that summarize a book's themes and reader reception alongside the traditional narrated sample. The feature targets audiobook discovery, where a short sample reveals the narrator but says little about whether a long book is worth the commitment.

Audible 用 AI 生成音频预览,帮用户决定一本书值不值得听
Read details: US Tightens AI Chip Export Controls to Cover Remote Access to Overseas Compute
Policy / Export Controls / Nvidia

US Tightens AI Chip Export Controls to Cover Remote Access to Overseas Compute

The Commerce Department is rolling out an interim final rule requiring licenses to send AI chips and servers to data centers anywhere in the world if that compute could reach companies in China or other restricted countries. It sets a 12 million TPP threshold, roughly 750 Nvidia H100 or H200 chips, and rescinds the Biden-era AI Diffusion Rule.

美国收紧AI芯片出口管制,远程调用海外算力也要许可证
Read details: OpenAI Reportedly Disbands Standalone Robotics Team, Folding Researchers Into Core Work
OpenAI / Robotics / Organization

OpenAI Reportedly Disbands Standalone Robotics Team, Folding Researchers Into Core Work

According to The Rundown AI, OpenAI has dissolved its stand-alone robotics team and moved its researchers into core research and related groups instead of keeping a dedicated embodied-AI unit. Coverage frames the move as a bet that general-purpose models can increasingly run robots without a dedicated robotics department.

OpenAI 被曝解散独立机器人团队,研究人员并入核心研究
Read details: Report: OpenAI explored a $5B-plus deal for Goodfire at 4x its last valuation
OpenAI / Acquisition / Business

Report: OpenAI explored a $5B-plus deal for Goodfire at 4x its last valuation

OpenAI explored acquiring the startup Goodfire for at least $5 billion, according to a November 16 report from FourWeekMBA, roughly four times the company's last valuation. The report describes explored talks rather than a signed agreement, and neither OpenAI nor Goodfire has publicly confirmed the discussions.

报道称OpenAI曾洽谈以至少50亿美元收购Goodfire,估值约为上一轮四倍
Read details: Uber's Legal AI Agent Cuts Contract Review Time 20% With Rules and RAG
Uber / AI Agent / Enterprise

Uber's Legal AI Agent Cuts Contract Review Time 20% With Rules and RAG

InfoQ reports that Uber's legal AI agent reduced contract redlining time by about 20% by pairing a deterministic rules layer with retrieval-augmented generation. The combination keeps the agent's suggestions anchored in the company's own clauses and legal standards rather than free-form model output.

Read details: Freshworks alums raise $53M to take on Salesforce and HubSpot with AI-native CRM
Dreamteam / AI CRM / Funding

Freshworks alums raise $53M to take on Salesforce and HubSpot with AI-native CRM

Dreamteam, a San Francisco startup founded by former Freshworks executives, has raised $53 million in seed and Series A funding led by Together Fund and Accel to challenge Salesforce and HubSpot with an AI-native CRM. Its Company Brain and built-in AI agents capture customer context automatically, and the company says it has signed more than 100 customers across 13 countries in three months of selling.

Freshworks 前高管创立 Dreamteam,融资 5300 万美元以 AI 原生 CRM 挑战 Salesforce
Read details: McKinsey: Banks could recover AI agent investments in about 10 months
McKinsey / AI Agent / Banking

McKinsey: Banks could recover AI agent investments in about 10 months

A new McKinsey report estimates that banks deploying AI agents could recoup their investment in roughly 10 months and earn nearly 2.9 times their initial outlay over three years. Drawing on 52 agentic AI use cases in banking, the firm argues the payoff hinges on workflow redesign, data quality and governance rather than layering agents onto existing processes.

Read details: Shanghai Maritime University unveils maritime law restatement agent that makes shipping contract rules traceable
上海海事大学 / 法律AI / 智能体

Shanghai Maritime University unveils maritime law restatement agent that makes shipping contract rules traceable

Shanghai Maritime University has presented an “International Maritime Law Restatement Agent” that turns sprawling rules for interpreting international shipping contracts into a systematic, traceable knowledge network for codification, adjudication and arbitration work. Built on the university's own maritime law model, it pushes vertical legal AI from question answering toward checkable professional workflows.

Read details: ifanr tests Tencent's Hy4 flash: a 3D interactive webpage for about 0.2 yuan
Tencent / Hy4 flash / Coding

ifanr tests Tencent's Hy4 flash: a 3D interactive webpage for about 0.2 yuan

Chinese tech outlet ifanr has published a hands-on test in which Tencent's Hy4 flash built a 3D interactive webpage from scratch for roughly 0.2 yuan. The piece frames the result around cost, asking what a near-zero bill for a full attempt means for everyday web development.

Read details: OpenAI case study: Revolut cuts coding time 30–60% with Codex
OpenAI / Codex / Enterprise

OpenAI case study: Revolut cuts coding time 30–60% with Codex

OpenAI has published a customer story on Revolut, saying the fintech uses its models and Codex to cut coding time by 30–60% and to scale financial crime investigations. The case study argues Revolut expanded its ambitions without lowering its security bar, though the numbers come from the vendor itself.

Read details: LVMH's media division announces strategic partnership with OpenAI
OpenAI / LVMH / Media

LVMH's media division announces strategic partnership with OpenAI

LVMH's media division and OpenAI have announced a strategic partnership aimed at supporting journalism with AI and making the division's publications more accessible through ChatGPT. The announcement sets out those two goals but does not disclose financial terms or which publications are covered.

LVMH 媒体部门与 OpenAI 达成战略合作,让旗下出版物通过 ChatGPT 更易获取
Read details: Nokia leads five-company consortium bidding for EU-funded Nordic-Baltic AI gigafactory
Nokia / Infrastructure / Europe

Nokia leads five-company consortium bidding for EU-funded Nordic-Baltic AI gigafactory

Nokia is leading a five-company consortium bidding for an EU-funded AI gigafactory project in the Nordic-Baltic region, according to Crypto Briefing. The report identifies only the five-firm grouping and Nokia's role as lead, without naming the other members, a bid value, or a decision timeline.

Read details: OpenAI Kicks Off First Phase of Embedded Assessments With Four Outside Research Groups
OpenAI / AI Safety / Evaluation

OpenAI Kicks Off First Phase of Embedded Assessments With Four Outside Research Groups

OpenAI has begun the first phase of its embedded assessments program with Apollo Research, METR, Transluce, and Goodfire, which will independently evaluate model alignment, safeguards, and monitoring systems. The work puts outside researchers inside the development process rather than only after release.

Read details: Amazon uncovers AI agent campaign stealing API keys from LiteLLM, Dify and other apps
Amazon / AI Agents / Security

Amazon uncovers AI agent campaign stealing API keys from LiteLLM, Dify and other apps

Amazon's threat intelligence team found attackers using AI agents to automatically hunt internet-exposed AI applications such as LiteLLM, One API and Dify in order to steal stored model provider API keys. The captured records show stolen keys being used to run the same models that powered the search for more keys.

Read details: Tencent Releases Hy4 Flash, a 224B Model Priced at 2 Yuan per Million Output Tokens
Tencent / Hunyuan / Model Release

Tencent Releases Hy4 Flash, a 224B Model Priced at 2 Yuan per Million Output Tokens

Tencent launched Hy4 flash on November 16, a 224-billion-parameter model priced at 2 yuan per million output tokens and pitched at everyday office work. The company says a single task costs roughly one-twentieth of the full Hy4 model, with access through WorkBuddy, Yuanbao and Tencent Cloud's TokenHub.

腾讯发布混元 Hy4 flash:224B 参数模型,输出每百万 token 定价 2 元
Read details: Zoho rebuilds Campaigns with unified email, SMS, WhatsApp and AI agent workflows
Zoho / AI Agent / Marketing

Zoho rebuilds Campaigns with unified email, SMS, WhatsApp and AI agent workflows

Zoho has rebuilt its Campaigns marketing platform, unifying email, SMS and WhatsApp in one workflow and adding native WhatsApp campaigns plus a new email editor. The platform supports the Model Context Protocol, so compatible AI agents can prepare campaigns, manage audience segments and review performance from natural-language instructions.

Zoho重构Campaigns平台:三渠道合一,接入MCP智能体工作流
Read details: OpenAI adds computer-use agents and 20-plus plugins to ChatGPT for Healthcare
OpenAI / Healthcare / Agent

OpenAI adds computer-use agents and 20-plus plugins to ChatGPT for Healthcare

OpenAI updated ChatGPT for Healthcare with computer-use agents that click and type across clinical software, improved clinical search and more than 20 new healthcare plugins. The company says the capabilities are available today in ChatGPT Work, Codex and the API, backed by results on a Stanford administrative benchmark and a safety review of EHR inbox tasks.

Read details: Weimob becomes authorized partner for Tencent's Workbuddy and Codebuddy
腾讯 / 微盟 / AI Agent

Weimob becomes authorized partner for Tencent's Workbuddy and Codebuddy

Weimob has become an officially authorized partner for Tencent's Workbuddy and Codebuddy AI products, according to a Sina Finance report. The announcement ties Tencent's agent tools to Weimob's merchant-services base, though the scope of the authorization, revenue terms and delivery commitments have not been disclosed.

Read details: TSMC plans to invest up to $1B in Anthropic and OpenAI
TSMC / OpenAI / Anthropic

TSMC plans to invest up to $1B in Anthropic and OpenAI

TSMC plans to invest up to $1 billion in Anthropic and OpenAI, according to a Semiecosystem report. The reported move would tie leading-edge chip capacity more closely to the two frontier model developers, though the investment structure, split and confirmation status remain undisclosed.

台积电拟向 Anthropic 与 OpenAI 投资至多 10 亿美元
Read details: Alibaba's Qwen4-Max debuts with 2.4T total parameters and 71B active, halving token use
Alibaba / Qwen / Model Release

Alibaba's Qwen4-Max debuts with 2.4T total parameters and 71B active, halving token use

Alibaba's Qwen team has released Qwen4-Max, a model with 2.4 trillion total parameters and 71B active per inference pass, with reports saying it consumes about half the tokens of its predecessor. Coverage also claims a 62% lower cost per task than Qwen3.8-Max while outperforming it.

阿里千问发布Qwen4-Max:总参数2.4万亿、激活71B,Token消耗减半
Read details: Apple launches a study app with OpenEvidence to test AI answers on health questions
Apple / AI Health / OpenEvidence

Apple launches a study app with OpenEvidence to test AI answers on health questions

Apple has launched a standalone app for a research study on how people use AI for health, built with healthcare AI company OpenEvidence, so participants can ask everyday medical questions and get answers grounded in clinical literature. The study is open to U.S. users over 21 on iPhone 15 or later with Apple Intelligence enabled, and Apple aims to enroll up to 200,000 people for more than a year.

苹果联手OpenEvidence推出健康AI研究应用,计划招募20万用户
Read details: Anthropic's record IPO faces big questions before it reaches the market
Anthropic / IPO / AI Business

Anthropic's record IPO faces big questions before it reaches the market

France 24 reports that Anthropic's record-breaking IPO is heading toward the public markets while still facing a set of big questions. Rather than treating the listing as a settled triumph, the report frames the offering as a test of what a leading AI company must prove once its valuation is priced daily by public investors.

Anthropic“创纪录IPO”登场前,France 24列出待解难题
Read details: AI Day Seoul wraps with 2,000-plus attendees focused on physical AI and agents
NVIDIA / Physical AI / AI Agent

AI Day Seoul wraps with 2,000-plus attendees focused on physical AI and agents

Nvidia's event recap says AI Day Seoul drew more than 2,000 attendees to COEX over two days, bringing together Korean industry leaders, startups and developer communities around physical AI and AI agents. At the plenary, Nvidia robotics and edge AI vice president Deepu Talla said Korea is punching above its weight in the field.

AI Day Seoul 收官:2000余人齐聚首尔,物理AI与智能体成主线
Read details: The King's Trust and OpenAI launch £1m partnership to reach 25,000 young people
OpenAI / Partnership / AI Skills

The King's Trust and OpenAI launch £1m partnership to reach 25,000 young people

The King's Trust and OpenAI have launched a £1 million partnership intended to help up to 25,000 young people across the UK build practical AI skills for work. The announcement fixes both the funding figure and a reach target, putting the emphasis on employability rather than on product news.

Read details: Hugging Face hands model routing to OpenRouter to make open weights turnkey
Hugging Face / OpenRouter / Inference

Hugging Face hands model routing to OpenRouter to make open weights turnkey

Hugging Face has switched its Inference Providers service from a custom router to OpenRouter, which picks a provider for each call based on live price, speed and reliability and retries elsewhere on failure, with no code changes for developers. The two companies describe the move as the first step of a partnership aimed at making open-weight models as easy to call as closed APIs.

Hugging Face 把推理路由交给 OpenRouter:开源权重模型开箱即用
Read details: Indian chip startup Bharat Semi raises $24M seed to target global AI sensor market
Bharat Semi / Semiconductor / Funding

Indian chip startup Bharat Semi raises $24M seed to target global AI sensor market

Bharat Semi, an Indian chip startup, has raised a $24 million seed round as it targets the global market for AI-enabled sensors. The company says it has secured design wins for its AI sensor technology with customers including General Atomics.

印度芯片初创 Bharat Semi 完成 2400 万美元种子轮,瞄准全球 AI 传感器市场
Read details: OpenAI launches a new blog dedicated to science
OpenAI / AI for Science

OpenAI launches a new blog dedicated to science

OpenAI has published a post titled "Our new blog for science" on its news site, introducing a blog focused on scientific work. According to the description, it will share new findings, practical methods, and firsthand accounts from researchers using AI to advance scientific discovery.

Daily Archive

Daily Archive

Full Archive
2026-11-18Daily AI Archive | 2026-11-18This issue collects 50 AI updates from 2026-11-18, led by Google makes Tasks in Gemini free in six countries and deepens Workspace integration, Google to keep Gemini chats for two months, up from 72 hours, even with history off, Voyage Labs raises $12M to build Atlas, a personal AI agent for experiencing the world, Daydream's New AI Agent Acts as a Personal Stylist — and Shops for You.2026-11-17Daily AI Archive | 2026-11-17This issue collects 40 AI updates from 2026-11-17, led by Lambda Adopts NVIDIA Photonics CPO Switches to Harden 10,000 GPU AI Factories, Report: OpenAI Explored Buying Interpretability Startup Goodfire for Over $5 Billion, NVIDIA Extends Slurm to Put Quantum Processors Inside Supercomputing Workflows, Nvidia's Vera Rubin debuts in MLPerf with a training sweep and 10x agentic efficiency gain.2026-11-16Daily AI Archive | 2026-11-16This issue collects 35 AI updates from 2026-11-16, led by ByteDance opens cheap Seedance 2.5 Flash API, with a 720p ceiling, XRP Ledger logs 11.7 million sub dollar AI agent payments, t54.ai bets on settlement, OpenAI Launches ChatGPT Wallet for AI Agent Payments, 腾讯混元字节研究人员集结!北京AI智能体大会及个人Agent研讨会要来了 智东西.2026-11-15Daily AI Archive | 2026-11-15This issue collects 22 AI updates from 2026-11-15, led by OpenAI will let ChatGPT Health conversations feed model training from November 19, Asian governments move to lock in Nvidia chip supply after Jensen Huang's exit, DeepSeek Opens Chatbot Search, File and Voice Features to Developers, AWS Trainium Picks Up SGLang Support: 17 Models, Kimi K3 on Three Nodes.2026-11-14Daily AI Archive | 2026-11-14This issue collects 21 AI updates from 2026-11-14, led by ChatGPT adds a Wallet so AI agents can pay at checkout, Gemini 4 Hydrogen costs 76% less per task than 3.8 Flash, LexisNexis launches an AI agent detection tool, Cisco CEO says every employee inside the company will get a personal AI agent.2026-11-13Daily AI Archive | 2026-11-13This issue collects 36 AI updates from 2026-11-13, led by AMD Unveils New AI Data Center GPU, Shares Jump More Than 5%, After Huang's resignation, Trump floats Tim Cook to run Nvidia, Northern Trust and Microsoft deploy an AI agent that cuts onboarding prep from weeks to days, Bielik 11B v3.1 ships as a 56 language Polish open model.2026-11-12Daily AI Archive | 2026-11-12This issue collects 45 AI updates from 2026-11-12, led by OpenAI's quarterly revenue reportedly tops $10B for the first time, Simpro Launches JustDo AI Agent for Trade Businesses, Gopuff goes live inside Grok Bot with a fully integrated checkout, TechRadar Reports Grok Build Coding Agent Is Leaking Confidential Data.2026-11-11Daily AI Archive | 2026-11-11This issue collects 31 AI updates from 2026-11-11, led by Google Expands Its TPU Accelerator Family, Framed Around the AI Agent Era, RainFocus Launches Anthropic Connector Giving AI Agents Access to Event Data, NVIDIA Expands Open Source Commitment With OpenShell Donation and New AI Infrastructure, Nvidia CEO Jensen Huang’s affair with alleged Chinese spy: a timeline.2026-11-10Daily AI Archive | 2026-11-10This issue collects 39 AI updates from 2026-11-10, led by Northern Trust and Microsoft Deploy AI Agent, Cutting Client Onboarding Time by 83%, Anthropic Buys $3.4B of Natural Gas Turbines as AI Power Race Widens, Google Adds TPU 8tx and 8i Max to Its Gen 8 Lineup, Aimed at Agent Workloads, Medium's New TK App Puts AI Assistance Inside the Writing Process.2026-11-09Daily AI Archive | 2026-11-09This issue collects 29 AI updates from 2026-11-09, led by Cloudflare and Tools for Humanity Propose Identity Framework for AI Agents, Nvidia board opens independent review of Jensen Huang's ties to alleged Chinese spy, Report Says Anthropic Is Building AI Systems Twice as Fast as Last Year, GammaDX index finds only 12 of 50 Australian websites are agent ready.