8 articles
Moonshot AI released Kimi K2.7-Code, an open-source update to its K2 coding model line that uses the same trillion-parameter mixture-of-experts architecture as K2.6 and supports Op...
Google researchers introduced faithful uncertainty, a method that helps LLMs align responses with their internal confidence and offer qualified “best guesses” when appropriate. The...
Microsoft released SkillOpt, an MIT-licensed framework that treats AI agent skill documents as optimizable objects rather than static instructions. The system uses performance feed...
Xiaomi’s MiMo AI team released MiMo Code V0.1.0, an MIT-licensed, terminal-native AI coding assistant based on the OpenCode agent. The company says it outperformed Claude Code on l...
UC Berkeley RDI and more than 300 experts launched Agents’ Last Exam, a benchmark meant to test whether AI agents can complete long, economically valuable professional workflows. O...
Sapient researchers say their HRM-Text architecture trained a 1B-parameter foundation model from scratch using far less data and compute than standard Transformer-based LLMs. The m...
Apple’s WWDC 2026 updates recast Siri as a systemwide interface for enterprise app content, data, and actions across iPhone, iPad, Mac, Apple Watch, and Vision Pro. Developers can...
Cohere launched North Mini Code, an open-source 30B-parameter mixture-of-experts model for agentic software engineering that can run on a single HThe model supports a 256K-token co...
