Crusoe handed Google an unfinished 10-gigawatt site in Wyoming after the developer struggled to bring the project in on budget.
Gemini Omni 1.2 Flash reaches general availability at $0.07 per second for 720p, 30 percent below its predecessor.
Spaces turns a chat thread into a persistent project workspace that keeps updating from connected apps.
The seed round led by Inovia Capital funds a platform that turns messy real-world scans into physically accurate training worlds.
Tuesday's blue wave put two data center skeptics in charge of the state hosting the country's biggest AI buildouts.
Palo Alto Networks is said to be paying $500M for a three-year-old startup that guards the doors between agents and…
Companies are running more tokens than ever while paying almost nothing extra for them, and that is bad news for…
The lab's second generation ships as a 594B mixture of experts on Tinker, with open weights waiting on a safety…
A 31-million-parameter voice model runs offline in a retro room where a typewriter reads back whatever you write.
A persistent serving engine turns kernel tuning into a continuous loop, and Together says it beats the best published throughput…
Reuters reports roughly 600,000 of Nvidia's H200 accelerators are sitting in Hong Kong while Beijing declines to approve bulk shipments…
An experimental cyber model found and proved out 948 Linux root exploits in a single October run that cost $1.2M…
Baidu published the raw pre-training weights of ERNIE-5.5-Flash-Base on Hugging Face, a 280B mixture-of-experts checkpoint that has never been taught…
Adweek reports Anthropic paid the Associated Press for election data that surfaced in Claude, the lab's first deal with a…
A cryptographic fingerprint could let users check that the model they paid for is the model that replied, without labs…
Reverse game theory gives up on reading a model's mind and tries to rig the incentives around it, an approach…
The venture firm is closing its London office and ending its Arc and Scouts programs, betting that fewer, more senior investors serve founders better.
An open model from the WeChat vision team handles detection, counting, OCR and pixel masking without bolting on a separate segmentation stack.
Serverless model hosting has turned into one of the most expensive races in AI infrastructure,…
Ballot measures and midterm upsets turned server farms into one of the most punishing issues…
The text detector is moving into visual forensics with a flagship image model and a…
MIT researchers say GPT-6 Astra changed 0.016 percent of Qwen's parameters, taking MATH-500 from 7.9…
Google's threat team says stolen model access resold through underground gateways produced single-day bills above…
The NCSC judges it highly likely that AI-powered malware features in a ransomware attack on…