This issue collects 23 AI updates from 2026-07-16, led by 美国AI新秀首发模型，参考中国技术|DeepSeek V3|Kimi K2.5|OpenAI|米拉·穆拉蒂|金融时报 手机新浪网 新浪财经, Applied Computing wants to give oil and gas operators an AI model for the entire plant, NVIDIA Introduces New Jetson Thor Computers to Advance Mainstream Robotics and Edge AI, Agentic orchestration: Enterprise AI organizations have a deployment problem, not a platform problem — and most are calling chatbots agents.
美国AI新秀首发模型，参考中国技术|DeepSeek V3|Kimi K2.5|OpenAI|米拉·穆拉蒂|金融时报 手机新浪网 新浪财经.5|OpenAI|米拉·穆拉蒂|金融时报 手机新浪网 新浪财经 Watch real task performance, API pricing, context capability, and migration cost.
Applied Computing wants to give oil and gas operators an AI model for the entire plant.Applied Computing has raised a $20M Series A to build a foundation AI model for the oil, gas and petrochemical industry.
Model Routing Is Simple.Until It Isn’t.Watch real task performance, API pricing, context capability, and migration cost.
谷歌升级 Gemini Spark AI助手，新增 Workspace 编辑能力并提速50%以上 article.9466.com.Watch real task performance, API pricing, context capability, and migration cost.
Apple Intelligence has received regulatory approval to launch in China through partnerships with Alibaba's Qwen AI and Baidu.The deal, rumored since last year, marks a critical step for Apple's AI ambitions in its most important overseas market.
Galaxy General has unveiled WAM TTT, the world's first test time training framework for embodied intelligence large models, bringing the TTT paradigm from NLP to physical robot control.The framework enables robots to adapt to new scenes using only unlabeled human videos, eliminating the need for expensive teleoperation data.
A hacker gained access to Suno's source code through a supply chain attack, revealing evidence that the AI music generator scraped decades of audio from YouTube Music, Deezer, and other platforms.The breach exposes Suno's training data practices at a time when the company faces copyright lawsuits from major record labels.
New details in Apple's trade secret lawsuit against OpenAI reveal a former Apple engineer who retained network access after joining OpenAI exploited a zero day bug to download dozens of confidential hardware files.Apple also alleges OpenAI's hardware chief asked candidates to bring Apple's unannounced product components to interviews.
NVIDIA Introduces New Jetson Thor Computers to Advance Mainstream Robotics and Edge AI.General purpose robots and autonomous machines are moving from research labs to real world mass market deployment, creating demand for compact, power efficient AI supercomputers capable of running foundation models at the edge.To meet that need, NVIDIA today.
Thinking Machines amps up its bet against one size fits all AI with its first open model, Inkling.It's the company's first public proof point after a year and a half spent building AI infrastructure largely out of public view.
Zhejiang Province has released a high quality dataset co creation platform for the cultural tourism industry, aimed at accelerating AI data infrastructure development.The initiative represents a significant push to standardize and share tourism data for AI applications.
Data center operator TeraWulf has signed a $19 billion lease agreement with AI company Anthropic, putting its brownfield AI infrastructure strategy to the test.The massive deal underscores the insatiable demand for compute capacity among frontier AI labs.
Chinese AI company MiniMax has globally launched its next generation M3 model, featuring compute costs as low as a few cents per day.The model also demonstrates AI stock trading capabilities and next generation interactive video generation.
Agentic orchestration: Enterprise AI organizations have a deployment problem, not a platform problem — and most are calling chatbots agents.Across 101 enterprises, agent orchestration is consolidating onto model provider platforms — Anthropic’s Claude leads by a wide margin — chosen for the gravity of the underlying model and judged on reliable multi step execution.But the ambition runs well ahea.
Meet GPT Red: an LLM super hacker OpenAI built to make its models safer.OpenAI has built an LLM super hacker called GPT Red that it uses as a sparring partner to help its other models boost their defenses against cyberattacks.Last week the company released the latest version of its flagship LLM, GPT 5.
Microsoft is training its sales force to position its own AI models as more efficient and cost effective than OpenAI and Anthropic's offerings, according to TechCrunch.The strategy highlights the increasingly complex relationship between Microsoft and the two AI companies it has invested heavily in.
China Mobile's chain fund has made a strategic investment in industrial AI company Youji Technology, as the competition in industrial AI shifts from model capability to industrial system understanding.Youji specializes in transforming traditional machine tools into intelligent systems with sensing, learning, and optimization capabilities.
Livestream shopping platform Whatnot has acquired Shaped, a machine learning company specializing in real time recommendations and search.The deal aims to solve live commerce's toughest challenge: helping shoppers find products as inventory, auctions, and buyer demand change by the second.
Amid hardware legal battle, OpenAI releases a $230 keyboard for Codex.OpenAI, which is in the middle of a legal battle with Apple over hardware trade theft allegations, just released a light up keyboard designed to be paired with its agentic coding app.
AI Skill 和 AI Agent，到底差在哪？53AI.AI Skill 和 AI Agent，到底差在哪？53AI Watch planning quality, tool use, permission control, and result verification.
Visa has warned that the surge of AI agent initiated micropayments often under $0.01 per transaction could overwhelm existing global credit card infrastructure within years.The company projects up to 80 billion AI agent micropayments per day by 2030 and is urging the industry to develop new batch settlement and offline clearing standards.
Demis Hassabis, CEO of Google DeepMind, urged the U.S.government to establish a national AI model testing and evaluation framework, similar to how the FDA approves drugs or the FAA certifies aircraft.
Chinese AI lab Moonshot AI's upcoming Kimi K3 model is expected to match or surpass Anthropic's Opus 4.8 in performance, the Financial Times reported.With 2 3 trillion parameters, it would be the largest open weight AI model from China.
