MiniMax Sparse Attention (MSA): a Two-Branch Block-Sparse Attention Trained on a 109B-Parameter MoE With a 3T-Token Budget

MiniMax released MSA (MiniMax Sparse Attention), a sparse attention method built directly on Grouped Query Attention (GQA). It targets one bottleneck: the quadratic cost of softmax attention at long context. The MiniMax research team tested it inside a 109B-parameter Mixture-of-Experts model trained with native multimodal data. They also open-sourced an inference kernel and shipped a…

Read Full News

Pinterest launches an experimental AI shopping app called ‘Ask Pinterest’

Pinterest on Wednesday announced a new experimental app called “Ask Pinterest” that will allow the company to explore a more conversational approach to shopping and product discovery that could eventually find its way to the main Pinterest app. It also introduced other AI initiatives, including Pinterest Model Context Protocol (MCP), designed for advertisers running campaigns…

Read Full News
Plaud says its software program enterprise topped 0M in ARR after delivery over 2M AI notetakers

Plaud says its software program enterprise topped $100M in ARR after delivery over 2M AI notetakers

There aren’t many success tales to check with with regards to AI {hardware}. Plaud, which makes AI-powered notetakers, is making an attempt to turn into one by focusing on professionals who take loads of conferences. The corporate mentioned it has bought greater than 2 million of its gadgets, together with Plaud Pins and credit-card-styled devices…

Read Full News
DOJ claims xAI’s unpermitted gasoline generators are a matter of ‘nationwide, financial, and vitality safety’

DOJ claims xAI’s unpermitted gasoline generators are a matter of ‘nationwide, financial, and vitality safety’

The Division of Justice on Monday sided with xAI in a lawsuit that sought to cease the corporate’s use of dozens of unpermitted pure gasoline generators close to its Memphis knowledge facilities, based on Wired. The DOJ stated if the NAACP, which filed the lawsuit in April, prevails, the end result would undermine “American nationwide,…

Read Full News

OpenAI’s Deployment Simulation Extends Pre-Deployment Risk Assessment to Agentic Coding Through Simulated Tool Calls

OpenAI published a new pre-deployment safety method called Deployment Simulation. The idea is direct. Before a model ships, simulate its deployment first. Replay past conversations through the new candidate model. Then study how it behaves in realistic contexts. OpenAI already uses insights from the method during model development. It has…

Read Full News