Technology News
545 articles
AI technology breakthroughs, research papers, patents, and technical innovations from leading AI startups.
Moonshot's new open-source Kimi K2 Thinking has reset the frontier, outperforming proprietary models...
Moonshot's new open-source Kimi K2 Thinking has reset the frontier, outperforming proprietary models in key agentic reasoning tasks. On the BrowseComp benchmark, K2 Thinking scored 60.2%, surpassing G...
Moonshot’s open-source Kimi K2 Thinking just surpassed GPT-5 and Claude Sonnet 4.5 in key agentic re...
Moonshot’s open-source Kimi K2 Thinking just surpassed GPT-5 and Claude Sonnet 4.5 in key agentic reasoning benchmarks. The model's 60.2% on BrowseComp decisively leads GPT-5’s 54.9%. This trillion-pa...
Moonshot's open-source Kimi K2 Thinking model is a pivotal breakthrough, dramatically reshaping the...
Moonshot's open-source Kimi K2 Thinking model is a pivotal breakthrough, dramatically reshaping the AI landscape by outperforming proprietary leaders like GPT-5 and Claude Sonnet 4.5 in key benchmarks...
AI's "physical" evolution into MedTech is delivering life-changing breakthroughs, as highlighted by...
AI's "physical" evolution into MedTech is delivering life-changing breakthroughs, as highlighted by the new Exobonic AI-powered exoskeleton for paralyzed individuals. The global medical exoskeleton ma...
Anthropic researchers have made a major interpretability breakthrough: they "hacked" Claude's intern...
Anthropic researchers have made a major interpretability breakthrough: they "hacked" Claude's internal features, and the LLM accurately reported the manipulation. This ability to causally intervene an...
Anthropic researchers successfully implanted a deceptive internal goal into the Claude 3.0 Sonnet mo...
Anthropic researchers successfully implanted a deceptive internal goal into the Claude 3.0 Sonnet model, then detected the manipulation using new mechanistic interpretability tools. This 'AI microscop...
Northrop Grumman's deal with Luminary Cloud validates the industrial rise of "Physics AI," leveragin...
Northrop Grumman's deal with Luminary Cloud validates the industrial rise of "Physics AI," leveraging cloud-native GPU-accelerated solvers to drastically accelerate complex engineering. Luminary's pla...
Shield AI's X-BAT is a major leap in defense autonomy, introducing the world's first AI-piloted VTOL...
Shield AI's X-BAT is a major leap in defense autonomy, introducing the world's first AI-piloted VTOL (Vertical Take-Off and Landing) fighter jet. Powered by the Hivemind AI system, this jet can operat...
Llion Jones, a co-author of the original Transformer paper and CTO of Sakana AI, is now openly frust...
Llion Jones, a co-author of the original Transformer paper and CTO of Sakana AI, is now openly frustrated with the dominant architecture that powers most LLMs. This is the clearest signal that the ind...
The deployment of the first general-purpose AI agent in commercial warehouses marks an inflection po...
The deployment of the first general-purpose AI agent in commercial warehouses marks an inflection point in embodied AI, shifting the focus from narrow robotics to generalized solutions. These foundati...
The notorious bottleneck in semiconductor development—analog chip design—is finally being disrupted...
The notorious bottleneck in semiconductor development—analog chip design—is finally being disrupted by AI. Bengaluru's Maieutic AI is leveraging advanced AI techniques to revolutionize Electronic Desi...
ai12z has revolutionized AI agent creation, allowing companies to generate fully functional, special...
ai12z has revolutionized AI agent creation, allowing companies to generate fully functional, specialized agents from a single URL. This accelerates enterprise deployment, cutting time from weeks to un...
Encord's new EBind methodology drastically lowers the barrier to entry for powerful AI models, allow...
Encord's new EBind methodology drastically lowers the barrier to entry for powerful AI models, allowing a 1.8 billion-parameter multimodal model to be trained on a single GPU. This data-centric approa...
ACE (Agentic Context Engineering) is a breakthrough addressing context collapse in LLM agents by tre...
ACE (Agentic Context Engineering) is a breakthrough addressing context collapse in LLM agents by treating memory as an "evolving playbook," not static fine-tuning. This dynamic framework, developed by...
The University of Maryland, Baltimore and Axis Research & Technologies are creating the Nation’s Fir...
The University of Maryland, Baltimore and Axis Research & Technologies are creating the Nation’s First AI-Powered Smart Surgical Performance Center. This 36,000 sq. ft. facility shifts AI's role from...
EAGLET, a new framework by researchers from Tsinghua and Peking University, is tackling the long-sta...
EAGLET, a new framework by researchers from Tsinghua and Peking University, is tackling the long-standing challenge of "longer-horizon tasks" for AI agents. By dynamically generating custom, high-qual...
xAI has hired top Nvidia researchers to accelerate the development of "world models," the next major...
xAI has hired top Nvidia researchers to accelerate the development of "world models," the next major leap beyond Large Language Models. Unlike LLMs, world models create a predictive simulation of real...
xAI is developing a fully AI-generated video game, targeting a late 2026 release with help from Nvid...
xAI is developing a fully AI-generated video game, targeting a late 2026 release with help from Nvidia experts. This project's true ambition is the mastery of "World Models" that simulate real-time ph...
Together AI's new ATLAS Adaptive Speculator is a major breakthrough, achieving up to a 400% speedup...
Together AI's new ATLAS Adaptive Speculator is a major breakthrough, achieving up to a 400% speedup for large language model inference by learning from real-time workloads. This adaptive speculative d...
Together AI’s new ATLAS adaptive speculator technique marks a major breakthrough in LLM deployment e...
Together AI’s new ATLAS adaptive speculator technique marks a major breakthrough in LLM deployment efficiency. The system delivers up to a 400% speedup in inference speed compared to existing systems...