Explore the Newsverum news archive, editorial policies and categories in Turkish and English, with Arabic, French, Spanish, German and Portuguese editions. (Page 264 of 436)
Google.org awarded a UC San Diego team $2.35 million to improve climate models and guide new ocean observations. The team plans to use AI agents alongside physics-based models and observational data.
Oak Ridge National Laboratory introduced Motor Decide on October 13. The platform aims to help manufacturers compare electric motor technologies by considering performance, material availability, supply-chain stability and technological maturity together.
Oak Ridge National Laboratory1 min read Data included
Oak Ridge National Laboratory reported on October 13 that a manufacturing method developed with its partners for Type V gas storage tanks won the ACE equipment and tooling innovation award. Infinite Composites, JuggerBot 3D and Oklahoma State University participate in the project.
Oak Ridge National Laboratory1 min read Data included
Anthropic’s Frontier Red Team assessed whether small open-weight models could establish inference on unfamiliar consumer hardware and coordinate autonomous spread in controlled tests. The company reports that a 9-billion-parameter model ran on eight of nine tested hardware types, while models at 30 billion parameters or below could perform some component tasks.
Anthropic Frontier Red Team1 min read Data included
Meta combines locally idle pockets across regions and accelerator types into a shared, opportunistic inference pool, even while the fleet is constrained overall. Its ads-ranking system uses two elasticity levers: Rank More Ads evaluates additional candidates, while Serve Better Models changes the share of requests sent to a more powerful model.
Cloudflare announced Clef-omni, an open-weight decision model that accepts audio and video alongside text and images. Instead of chaining separate speech transcription and video-frame models, it classifies and selects field values from multiple modalities within one request.
Cohere says the choice between shared and dedicated inference for Embed and Rerank depends on work per request, traffic patterns, utilization and latency targets. Requests per minute alone can mislead: batch catalog indexing is throughput-oriented, while interactive search needs low latency.