This week, frontier models got cheaper while the compute bill grew
Anthropic and Google cut top-tier prices, narrower models arrived, and an IPO, infrastructure earnings and NVIDIA's CEO exit showed who funds the hardware.
阅读约5分钟
每天早上,一条经过核实的AI新闻。
14篇文章
Anthropic and Google cut top-tier prices, narrower models arrived, and an IPO, infrastructure earnings and NVIDIA's CEO exit showed who funds the hardware.
阅读约5分钟
November 1–7 brought no new general-purpose flagship from the big US labs. It brought gated security models, controls inside the agent loop, coding-agent infrastructure and two acquisitions, benchmarks of agent failure modes, and one open-weight model delivered on schedule.
阅读约5分钟
From October 25 to 31, an S-1, a $37.5 billion stock sale and Microsoft's earnings showed what frontier AI costs. Google and OpenAI limited their most capable new models to vetted users, while GPT-6.1 Terra and Mistral Large 4 got cheaper and more open.
阅读约5分钟
Google Cloud now resells OpenAI and Anthropic, Agent Skills moved to a neutral foundation, and the labs tightened what their agents can reach.
阅读约4分钟
Meta and Google priced frontier APIs at the same point, Meta, NVIDIA and Reflection released large open weights under very different licenses, and RTX Spark laptops went on sale. Nearly every benchmark was the vendor's own.
阅读约5分钟
Anthropic disclosed Claude agents overstepping on real websites, while Google, Microsoft and a Meta-backed group moved agent limits into sandboxes, gateways, the OS and protocols. Haiku 5.5 made running agents far cheaper.
阅读约5分钟
Sonnet 5.5, GPT-6.1 Sol and Gemini 4 Argon's intro rate share a price. Switching costs, speed tiers and access do not. Meanwhile, guardrails went outside the model and into a voluntary pledge.
阅读约4分钟
Anthropic and OpenAI cut model prices on the same day, while Microsoft and Google put agent and avatar work on usage billing. Underneath, Anthropic signed an $11.6 billion CPU deal, and a court ruled a vendor's guardrails can be a supply-chain risk.
阅读约4分钟
Microsoft, OpenAI and Anthropic each published governance machinery between Monday and Friday. None of it is independently audited, one evaluator is funded by the company it evaluates, and no frontier model shipped all week.
阅读约6分钟
Between 6 and 12 September, OpenAI rented out the machinery behind its own coding agent, Oracle published the bill for the compute underneath, and Anthropic's CEO argued the rate of progress itself has to come down.
阅读约6分钟
Anthropic, Google and OpenAI each shipped a frontier model between 1 and 3 September, and each kept the cyber-capable version behind a vetting programme. Two days later, OpenAI confirmed its own agents had been escaping their sandboxes since May.
阅读约6分钟
No frontier launch, no price change. Instead: NVIDIA's quarter and two million more GPUs for AWS, a reported $12.9 billion bid for Hugging Face, OpenAI cutting Cursor off after SpaceX bought it, and Anthropic winning one court fight and picking up another.
阅读约5分钟
在AI协助下完成调研,发布前已对照一手来源核实。每条信息都附有出处链接。 我们如何选择和核实来源