OpenAI Ships Ultrafast Preview for GPT-5.6 Sol, Promising 750 Tokens Per Second
The new mode, powered by Cerebras chips, targets enterprise workflows where latency matters more than deliberation - but the rollout stays gated for now.
Frontier models, infrastructure, and the policy that shapes them. Reported across Seoul, Singapore, Bengaluru, and Hangzhou.
The new mode, powered by Cerebras chips, targets enterprise workflows where latency matters more than deliberation - but the rollout stays gated for now.
Cupertino's variable-compensation approach diverges sharply from OpenAI's upfront licensing playbook as the race for fresh training data int…
How a Beijing medical resident used OpenAI's latest model to crack a problem that had stumped mathematicians since 2004 - while researching …
The tech giant is consolidating its fragmented AI assistant strategy, cutting experimental features and admitting its approach lost clarity …
Meta's new Scam Alert uses local inference to flag suspicious chats without server-side data sharing, marking a shift in how messaging platf…
The Hangzhou-based AI lab's latest model introduces enhanced agent capabilities alongside a pricing shift that signals a new phase in China'…
The company's Qwen3.8-Max model remains free to download, but enterprises above $50 million in revenue now face commercial licensing require…
A coordinated push by insurance firms, investment banks, and provincial governments signals China's intent to dominate the emerging BCI sect…
Wuhan-based chipmaker captured 14% of Q2 bit shipments, marking Beijing's first breakthrough in a market long dominated by Korean and Japane…
The Chinese AI start-up's quiet refresh shows uneven progress, with benchmark scores trailing expectations while specialized cybersecurity p…
SMIC and Hua Hong post triple-digit profit jumps as Beijing's semiconductor push collides with US export policy, reshaping Asia's foundry la…
Facebook's new stand-alone application offers personalized growth recommendations and automated comment management as the platform battles f…
Two designers have built ShieldFont, a typeface that uses ligatures to feed meaningless text to training crawlers while keeping web pages pe…
A 40-minute compromise of an open-source AI toolchain exposed terabytes of access secrets from Microsoft, Amazon, Cisco, and thousands of ot…
Geoffrey Hinton, Fei-Fei Li, and Andrew Ng tackle the tension between democratizing AI and concentrating power in a handful of labs.…
Independent booksellers report unusual bulk purchases of rare and out-of-print volumes, followed by their likely destruction in the rush to …
The platform's new control lets streamers exclude VODs, clips, and chat from future generative model datasets, though enforcement questions …
A new survey of 300 data executives reveals that legacy infrastructure is starving enterprise AI agents of the information required to make …
Streamers can now block their content from feeding Amazon's generative models, but the default remains opt-in for material dating back to 20…
The streaming platform has added a toggle to let users block their streams, clips, and chat data from training Amazon's generative AI system…
The company now allows unlimited text conversations at no cost but restricts access to advanced models and introduces advertising to non-pay…
As autonomous vehicles scale across Chinese cities, a new class of desk-bound workers is emerging to handle edge cases and emergencies the A…
A survey of 107 large organizations reveals that most deploy three agent-control systems at once, prioritizing flexibility over any single m…
Taiwan's manufacturing giant says integrated supply chain and production muscle position it to expand share as modular platforms reshape the…
A venture arm linked to YMTC has invested in SOI Micro, a domestic player pursuing fully depleted silicon-on-insulator technology as Beijing…