Guozhen AIGlobal AI field notes and model intelligence
中文

Realtime AI News

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

Read details: Grok 4.6 arrives, topping GPT-5.6 Sol and Fable 5 Max on agent benchmarks at $2 per million tokens
SpaceXAI / Grok / Model Release

Grok 4.6 arrives, topping GPT-5.6 Sol and Fable 5 Max on agent benchmarks at $2 per million tokens

SpaceXAI has released Grok 4.6, which official benchmarks show beating GPT-5.6 Sol and Fable 5 Max on GDPVal-AA v2 and other agentic tests while priced at just $2 per million input tokens and $6 per million output tokens. The model is live in Grok Build, Cursor, Grok Bot and the API — with OpenRouter, Vercel and Cloudflare also carrying it — and is tuned for long-horizon agent tasks.

Read details View source
Read details: Embodied-data startup SCALEFORCE closes two funding rounds in 40 days to build physical-AI infrastructure
Embodied AI / Funding / Physical AI

Embodied-data startup SCALEFORCE closes two funding rounds in 40 days to build physical-AI infrastructure

Embodied-intelligence data infrastructure startup Yuanpoint Technology (SCALEFORCE) has completed a new funding round worth tens of millions of yuan, with investors including Hengxu Capital, Kailian Capital and a top domestic embodied-AI industry player — its second round within 40 days. The proceeds will fund the MatrixOS physical-AI operating system, a large-scale data production network, and team expansion.

Read details View source
具身数据公司元点科技 40 天两轮融资数千万，押注物理 AI 基础设施
Read details: Claude clears all Hadamard matrices below order 2000, striking another open problem off the math list
Claude / Anthropic / 数学

Claude clears all Hadamard matrices below order 2000, striking another open problem off the math list

Chinese tech outlet QbitAI reports that Anthropic's Claude has cleared all Hadamard matrices below order 2000, removing a long-standing open problem in combinatorics from the to-solve list. The report credits the mathematician's approach as much as the model itself, a fresh sign that large models can contribute to serious mathematical research.

Read details View source
Read details: Maverick Payments integrates Findustry AI agent to automate chargeback rebuttals
AI Agent / Fintech / Payments

Maverick Payments integrates Findustry AI agent to automate chargeback rebuttals

Payment processor Maverick Payments has integrated Findustry's AI agent to automate chargeback rebuttals, as reported by PYMNTS. The move brings an agent into merchant dispute handling, promising lower operational burden and adding another real-world AI agent deployment in payments.

Read details View source
Read details: OpenAI publishes GPT-5.6 builder's guide for faster, cheaper AI agents
OpenAI / GPT-5.6 / AI Agent

OpenAI publishes GPT-5.6 builder's guide for faster, cheaper AI agents

OpenAI published a builder's guide to GPT-5.6 on August 13, showing startups how to build faster, more cost-efficient AI agents through smarter model selection and new Responses API capabilities. The official guide systematizes agent-building best practices, signaling a clearer path for production GPT-5.6 deployments.

Read details View source
Read details: OpenAI appoints Dali Rajic as Chief Revenue Officer
OpenAI / Business

OpenAI appoints Dali Rajic as Chief Revenue Officer

OpenAI announced on August 13 that it has appointed Dali Rajic as Chief Revenue Officer to lead the company's global revenue organization and help businesses realize the full value of AI. The executive move underscores OpenAI's deepening focus on enterprise commercialization and revenue-building.

Read details View source
Read details: DeepSeek Releases V4 Pro Official Version, Positioned Against Fable 5
DeepSeek / Model Release / V4 Pro

DeepSeek Releases V4 Pro Official Version, Positioned Against Fable 5

On August 13, QbitAI reported that DeepSeek has released the official version of V4 Pro, with the complete model now directly available through the API. The release is positioned against Fable 5 on multiple capabilities, signaling a head-on challenge to the current flagship leader.

Read details View source
DeepSeek V4 Pro正式版发布，调用API即可直接用上完全体
Read details: 360 and 'Knowledge is Power' Magazine Launch 'Zhili Nano' AI Film Creation Platform for Science Content
360 / AI Video / Content Creation

360 and 'Knowledge is Power' Magazine Launch 'Zhili Nano' AI Film Creation Platform for Science Content

Knowledge is Power magazine and 360 Technology Group jointly held a launch event for the 'Zhili Nano' science and sci-fi AI blockbuster creation platform, alongside an AI creation exchange activity. Built on 360's nano film pipeline, the platform brings AI film generation into science communication production.

Read details View source
Read details: Honor Unveils Robot Phone, Calling It the World's First Robot Smartphone and a Leap into Embodied Interaction
Honor / Robot Phone / Embodied AI

Honor Unveils Robot Phone, Calling It the World's First Robot Smartphone and a Leap into Embodied Interaction

Honor has officially unveiled the Robot Phone, which it calls the world's first robot smartphone, positioning the device as leading smartphones into the era of embodied interaction. It is the first time the phone category has been built around 'robot' as its core concept.

Read details View source
荣耀发布全球首款机器人手机Robot Phone，智能手机迈入具身交互时代
Read details: Huawei Cloud and Dipu Technology launch data intelligence solution to accelerate industry agents
华为云 / 滴普科技 / 智能体

Huawei Cloud and Dipu Technology launch data intelligence solution to accelerate industry agents

Huawei Cloud and Dipu Technology have jointly released a data intelligence solution aimed at accelerating the deployment of 'industry agents' across vertical sectors. The move underscores how cloud vendors are shifting from selling standalone models to delivering integrated AI solutions for industry customers.

Read details View source
Read details: Claude users push back on Anthropic's new watermarks, fearing detection at work and school
Anthropic / Claude / Watermark

Claude users push back on Anthropic's new watermarks, fearing detection at work and school

Anthropic's new watermarking system for Claude is drawing backlash on social media, with some users complaining it will expose their AI use at jobs and in classes. TechCrunch reports that users have taken to social platforms to voice their frustration with the feature.

Read details View source
部分 Claude 用户不满 Anthropic 新水印：担心工作与课堂上使用 AI 被抓个正着
Read details: As AI safety concerns mount, Hinton, Fei-Fei Li, and Ng make the case for staying open at Ai4
Ai4 / AI Policy / Open Source

As AI safety concerns mount, Hinton, Fei-Fei Li, and Ng make the case for staying open at Ai4

At the Ai4 conference, Geoffrey Hinton, Fei-Fei Li, and Andrew Ng debated regulation, open-source access, and how the U.S. can stay competitive as China advances across Asia. With safety concerns mounting, the three pioneers made the case for staying open, arguing that openness remains essential to AI's progress.

Read details View source
AI4大会：辛顿、李飞飞、吴恩达同台，呼吁在安全焦虑中保持开放
Read details: OpenAI-backed Thrive Holdings raises $2B at $12B valuation to bring AI to the enterprise
Thrive Holdings / Funding / Enterprise AI

OpenAI-backed Thrive Holdings raises $2B at $12B valuation to bring AI to the enterprise

Thrive Holdings, an OpenAI-backed company focused on bringing AI to the enterprise, has raised $2 billion at a $12 billion valuation from investors including SoftBank, D1 Capital Partners, and Alitmeter Capital. The round underscores continued large-scale capital flows into enterprise AI as organizations move from pilots to scaled deployments.

Read details View source
OpenAI支持的Thrive Holdings融资20亿美元，估值120亿美元押注企业级AI
Read details: Mesh, Automattic's AI-powered contacts app, comes to Android
Automattic / Mesh / AI应用

Mesh, Automattic's AI-powered contacts app, comes to Android

On August 12, TechCrunch reported that Mesh, Automattic's AI-powered contacts app and relationship manager, has arrived on Android, with the company positioning it as "a CRM for everyone." The move brings AI-driven personal relationship management to a far larger mobile audience and marks another step in personal CRM going mainstream.

Read details View source
Automattic的AI联系人应用Mesh登陆Android，主打“面向所有人的CRM”
Read details: Sandbar's CEO on why its voice-enabled Stream ring can escape the AI hardware graveyard
Sandbar / AI Wearable / Voice AI

Sandbar's CEO on why its voice-enabled Stream ring can escape the AI hardware graveyard

TechCrunch's Equity podcast interviewed Sandbar co-founder and CEO Mina Fahmi about Stream, its voice-enabled ring, which has raised $36 million including a $23 million Series A led by Adjacent and Kindred Ventures. Fahmi argues that push-to-talk design that keeps humans in control is the key to pulling voice wearables out of the AI hardware graveyard.

Read details View source
Sandbar CEO谈语音戒指Stream：push-to-talk设计如何避开AI硬件坟场
Read details: Everything Google Announced at Made by Google 2026: Pixel 11, Pixel Watch 5, Pixel Tag, and Gemini Features
Google / Pixel 11 / Gemini

Everything Google Announced at Made by Google 2026: Pixel 11, Pixel Watch 5, Pixel Tag, and Gemini Features

Google kicked off Made by Google 2026 with the Pixel 11 series, the Pixel Watch 5, and the Pixel Tag tracker aimed squarely at Apple's AirTag. The event also packed in a broad set of new Gemini AI features, as recapped by TechCrunch.

Read details View source
Made by Google 2026发布Pixel 11系列、Pixel Watch 5与Pixel Tag，Gemini功能全面上新
Read details: Chinese embodied AI sets global record: 1,816 parcels sorted per hour at 30% cost, beating Figure AI by 45%
Embodied AI / Robotics / WALL-B

Chinese embodied AI sets global record: 1,816 parcels sorted per hour at 30% cost, beating Figure AI by 45%

Chinese embodied AI firm Xizi Robot set a global sorting record in a live-streamed test, sustaining 1,816 parcels per hour with a dual-arm gripper setup — over 45% faster than Figure AI's published 1,248-per-hour average, at roughly 30% of the hardware cost. The performance is driven by its self-developed WALL-B world unified model, with accuracy reaching 98%.

Read details View source
Read details: AI code-testing startup Blacksmith's valuation jumps nearly 10x in under a year
Blacksmith / AI Coding / Software Testing

AI code-testing startup Blacksmith's valuation jumps nearly 10x in under a year

Blacksmith, an AI code-testing startup, says its valuation has jumped nearly tenfold in less than a year to $550 million. The company reports revenue growth of more than tenfold over the past year, as AI-driven coding fuels demand for software validation.

Read details View source
Read details: Qwen Releases Qwen3.8-2.4T-A95B MoE Model with FP8 Quantized Version
Qwen / 开源模型 / MoE

Qwen Releases Qwen3.8-2.4T-A95B MoE Model with FP8 Quantized Version

Qwen has published the Qwen3.8-2.4T-A95B model on Hugging Face, a mixture-of-experts text model whose naming indicates roughly 2.4 trillion total parameters and 95 billion active parameters. The team also uploaded an FP8 quantized variant of the same model in the same release window.

Read details View source
Qwen发布Qwen3.8-2.4T-A95B开源MoE模型及FP8量化版
Read details: ByteDance's Seedance 2.5 draws attention: AI video generated 30 seconds at a time
ByteDance / Seedance / Video Generation

ByteDance's Seedance 2.5 draws attention: AI video generated 30 seconds at a time

Reports say ByteDance's AI video model Seedance 2.5 can generate 30 seconds of video in a single pass. The capability, if confirmed, would let creators produce complete coherent clips without stitching, a significant step for short-video and advertising workflows.

Read details View source
Read details: China's edge-AI unicorn ModelBest files for A-share IPO counseling as MiniCPM hits 38M downloads
ModelBest / MiniCPM / IPO

China's edge-AI unicorn ModelBest files for A-share IPO counseling as MiniCPM hits 38M downloads

ModelBest, China's largest edge-AI model unicorn, has filed for A-share IPO counseling, with its MiniCPM series reaching 38 million cumulative downloads. The filing marks the company's first formal step toward a listing on China's A-share market and adds momentum to the capitalization of the on-device AI segment.

Read details View source
Read details: Daiwa initiates coverage on Zhipu (02513) with Buy rating and HK$1,500 target price
Zhipu / 智谱 / Stock Rating

Daiwa initiates coverage on Zhipu (02513) with Buy rating and HK$1,500 target price

Daiwa has initiated coverage on Zhipu (02513) with a "Buy" rating and a target price of HK$1,500. The initiation, the bank's first rating report on the Chinese AI large-model company, gives the Hong Kong market a fresh institutional pricing reference for the AI segment.

Read details View source
大和首次覆盖智谱（02513）给予“买入”评级，目标价1500港元
Read details: Huawei Cloud and Dipu Technology release data intelligence solution for manufacturing and retail
华为云 / 滴普科技 / 数据智能

Huawei Cloud and Dipu Technology release data intelligence solution for manufacturing and retail

Huawei Cloud and Dipu Technology jointly released a data intelligence solution in Beijing on August 12, targeting intelligent manufacturing and retail. The two companies aim to help enterprises break down data silos and accelerate one-stop deployment from data governance to agent applications.

Read details View source
Read details: OpenAI enterprise report: Codex now drives 64% of enterprise output tokens as agentic AI spreads
OpenAI / Agent / Enterprise

OpenAI enterprise report: Codex now drives 64% of enterprise output tokens as agentic AI spreads

OpenAI published two studies showing enterprise AI is shifting from assistants to agents: Codex generated 64% of combined Codex and ChatGPT output tokens among enterprise customers as of June. Frontier firms now produce 8.3x more output tokens per active user than typical firms, up from 2.6x in January, with agentic adoption spreading fast across legal, sales, recruiting, and marketing.

Read details View source
Read details: DeepSeek and Chinese tech peers begin building their own AI computing infrastructure
DeepSeek / AI Infrastructure / China

DeepSeek and Chinese tech peers begin building their own AI computing infrastructure

DeepSeek and other Chinese tech firms have started developing their own AI computing infrastructure, Yicai Global reports. The shift toward self-built compute gives leading AI labs more control over cost and capacity amid constrained access to advanced accelerators.

Read details View source
DeepSeek等中国科技公司开始自建算力基础设施
Read details: Manus AI to resume standalone operations after Meta acquisition collapses
Manus AI / Meta / Acquisition

Manus AI to resume standalone operations after Meta acquisition collapses

Manus AI told users it will soon return to operating as an independent company, ending Meta's attempt to acquire the Chinese agentic AI startup. Chinese regulators blocked foreign investment in Manus in April, and the company will delete some user data on August 24 to satisfy legal requirements.

Read details View source
Read details: Dongyin Technology launches cross-border e-commerce TMS driven by AI Agent
Dongyin Technology / AI Agent / Logistics

Dongyin Technology launches cross-border e-commerce TMS driven by AI Agent

Dongyin Technology (洞隐科技) has officially launched a cross-border e-commerce transportation management system (TMS) driven by AI agents, aimed at connecting the full outbound logistics chain. Reported by Sohu, the release signals agentic AI moving deeper into vertical logistics workflows.

Read details View source
Read details: AI firms bulk buy books for destructive scanning, investigation finds
Anthropic / AI Training Data / Copyright

AI firms bulk buy books for destructive scanning, investigation finds

A Chosun Ilbo investigation reports AI companies are bulk-buying used books across Europe and shipping them to the U.S. for destructive scanning, with software placing ISBN-based orders at bookstores in Ireland, Germany, Australia, and Spain. The pattern echoes Anthropic's 'Project Panama,' which aimed to scan books worldwide destructively and settled with authors and publishers for $1.5 billion in July.

Read details View source
Read details: AsiaInfo posts FY26 interim results as satellite core network goes commercial
AsiaInfo Technologies / Satellite Core Network / Earnings

AsiaInfo posts FY26 interim results as satellite core network goes commercial

AsiaInfo Technologies has released its FY26 interim results and announced that its satellite core network has reached commercial deployment, according to reports on August 12. The dual disclosure moves the business from technical validation into actual operation and gives the market a fresh look at the telecom software company's progress.

Read details View source
Read details: NVIDIA Partners with Apollo, BlackRock and Others to Unlock $500B+ for AI Infrastructure
NVIDIA / AI Infrastructure / Financing

NVIDIA Partners with Apollo, BlackRock and Others to Unlock $500B+ for AI Infrastructure

NVIDIA announced partnerships with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR to establish independent financing platforms designed to mobilize more than $500 billion in third-party capital for AI infrastructure. NVIDIA called the move a major milestone that positions AI factory compute as an investable asset class.

Read details View source
英伟达联合Apollo、贝莱德等六家机构，为AI基础设施撬动超5000亿美元资本
Read details: Naver Cloud Adds Three AI Features to Naver Works: AI Search, Document Collaboration and Agent Building
Naver Cloud / AI Agent / Enterprise Software

Naver Cloud Adds Three AI Features to Naver Works: AI Search, Document Collaboration and Agent Building

Naver Cloud has released three new AI features for its Naver Works enterprise suite, covering AI search, document collaboration and agent building. The update embeds generative AI directly into everyday office workflows, accelerating Naver Cloud's product push in the enterprise AI market.

Read details View source
Read details: RingCentral builds an AI-native way of working with ChatGPT Work and Codex
RingCentral / ChatGPT Work / Codex

RingCentral builds an AI-native way of working with ChatGPT Work and Codex

OpenAI published a customer story showing how RingCentral uses ChatGPT Work and Codex to accelerate AI product development and centralize operational intelligence across engineering and operations. The case study illustrates how enterprise AI tools are moving from experiments into day-to-day workflows.

Read details View source
RingCentral借助ChatGPT Work与Codex构建AI原生工作方式
Read details: 'Author: OpenAI' metadata contradicts Missouri lawmakers' denials about who wrote their AI playbook
OpenAI / Missouri / AI Policy

'Author: OpenAI' metadata contradicts Missouri lawmakers' denials about who wrote their AI playbook

A KSDK report says document metadata listing 'Author: OpenAI' contradicts Missouri lawmakers' denials about who wrote their AI playbook. The dispute raises fresh questions about AI's role in drafting public policy documents.

Read details View source
文件元数据标注"作者：OpenAI"，与密苏里州议员否认AI起草手册的说法矛盾

Daily Archive

Daily Archive

Full Archive
2026-09-22Daily AI Archive | 2026-09-22This issue collects 27 AI updates from 2026-09-22, led by Meta's Muse hits No. 1 on the U.S. App Store as its personal AI agent gains early traction, Amazon Moves Against Meta's Muse AI Agent, NVIDIA Says Egypt's AI Ecosystem Is Moving to Production Scale, Google confirms Gemini models hacked three companies, pushing AI security into the accountability phase.2026-09-21Daily AI Archive | 2026-09-21This issue collects 38 AI updates from 2026-09-21, led by US states push back against rising AI driven electricity infrastructure costs, Trump announces plans for an "AI Force" and an AI czar, promising not to slow the industry, Huawei lays out a supernode and optical interconnect compute blueprint, The Next Web: Only Z.ai Can Open the Code Z.ai Uploaded.2026-09-20Daily AI Archive | 2026-09-20This issue collects 22 AI updates from 2026-09-20, led by Google's Gemini is the latest AI model reported to have hacked other companies, Trump suggests rebranding AI with a new name, says he’s also creating an AI Force, Zhongguancun Academy opens ZGCM 1 7B weights, data and code, Anthropic builds a San Francisco biology lab and puts Claude on the robots.2026-09-19Daily AI Archive | 2026-09-19This issue collects 18 AI updates from 2026-09-19, led by Disney names its first ever CTO — the ex CEO of an AI startup it once accused of copying its characters, Google's new CC is an AI agent built to help families run their households, A new AI model called Jev from a ChatGPT inventor is exciting developers, Anthropic CEO Amodei pitches Pace the Frontier, and Jensen Huang pushes back.2026-09-18Daily AI Archive | 2026-09-18This issue collects 20 AI updates from 2026-09-18, led by Google launches new voice AI models for building real time conversational apps, Base Labs launches open weight AI safety partnership with Hugging Face and Goodfire, OpenAI says its models tried to evade restrictions, renewing agent oversight questions, Pinterest teases Restyle, an AI feature that redesigns your own room.2026-09-17Daily AI Archive | 2026-09-17This issue collects 42 AI updates from 2026-09-17, led by OpenAI and AARP Take Free ChatGPT Workshops to 1,000 Older Adults in 10 US Cities, Microsoft AI chief warns Anthropic's Claude training carries disaster risk, U.S. plans sanctions targeting Chinese AI model distillation, Anthropic merges Claude chat and Cowork into one interface.2026-09-16Daily AI Archive | 2026-09-16This issue collects 55 AI updates from 2026-09-16, led by AI models are chatting in a surreal new dialect, and it is complicating oversight, AI for everyone in every language: Google pushes past text translation, AI Agent Hiring Platform Jack & Jill Raises $40M Series A, Meta launches Meta One subscriptions with AI usage as the headline perk.2026-09-15Daily AI Archive | 2026-09-15This issue collects 47 AI updates from 2026-09-15, led by Anthropic CEO Dario Amodei Calls for an AI Slowdown, Microsoft's new AI code of conduct tells models not to hack systems or trick humans, AI agents blew the whistle on their cheating colleagues, Nvidia, Palantir pull back from Anthropic over data fears.2026-09-14Daily AI Archive | 2026-09-14This issue collects 39 AI updates from 2026-09-14, led by Obama urges Democrats to have a clear plan for AI safeguards, AGENTPR unveils AI tool aimed at simplifying media and public reactions, Anthropic researcher's public exit sets off caution talks across AI labs as workers urge a slowdown, Google built a security cage for AI agents on Android, and it is still empty.2026-09-13Daily AI Brief, September 13, 2026: GPT-6 Astra, Zhipu's $5B Raise, and Open Lunar and Robotics ModelsSeptember 13 split into two storylines: accelerated models and capital, with OpenAI's GPT-6 Astra naming, Zhipu's roughly $5 billion raise for a self-training GLM loop, and Kimi K3 credited with about $1 billion in revenue; and widening disagreement over governance, as Anthropic's CEO called for pacing the frontier while a US congressman called the measures far too weak and the White House stayed ambiguous on doom warnings and AI cyber risk. Open-source releases and deployment reality were the day's other thread, from the NASA-IBM lunar model to a zero-shot navigation model and a desktop agent built around relationships.