1841 articles
The U.S. government launched the Genesis Mission, a Department of Energy initiative designed to speed up AGI development through scientific AI, massive computing power, and automated research systems. The program brings together national computing resources, datasets, and top research talent under one roof.
Claude Opus 4.5 delivers a 21 percentage point accuracy boost on the WeirdML benchmark while slashing costs by two-thirds. The upgrade represents the biggest performance leap in the Opus lineup to date.
Alibaba just rolled out its Quark AI glasses in two versions—the S1 at $536 and the G1 at $261. Both run on the company's Qwen AI system, throwing Alibaba into the ring with Meta, Xiaomi, and others battling for dominance in AI-powered eyewear.
An open-source Deep Search tool using Google ADK and Gemini 3 is gaining traction for autonomous multi-source research. The system uses recursive reasoning to plan, validate, and generate structured reports.
Harvard's longevity lab tested Biostate AI's K-Dense Beta, a multi-agent AI system that compressed aging research from years into weeks. The platform managed literature reviews, code execution, and experiment design while running continuous reliability checks.
Grok 4.1 Fast just hit a massive milestone on OpenRouter, crunching through 253 billion tokens in one day. The spike shows how quickly this AI model is catching on with users.
MongoDB just dropped a comprehensive AI resource hub packed with free, production-grade engineering guides. Developers can access tutorials covering vector search, agent memory systems, and data infrastructure fundamentals—all designed to help bridge the gap between prototyping and real-world deployment.
OpenAI launched ChatGPT for Teachers, a secure GPT-5.1 workspace with unlimited messages, file uploads, search tools, and connectors for verified U.S. educators—free through June 2027 as AI adoption accelerates in K–12 classrooms.
The rapid expansion of AI infrastructure is pushing electricity consumption to new highs. Power companies are warning that utility bills across the board may climb significantly as data centers demand more energy over the next few years.
Elon Musk announced that Grok Imagine now supports text-to-video generation with upgraded audio and 10-second video output. The update removes the need for base images and introduces new creation modes inside the app.
OpenAI just dropped GPT-5.1 Instant and GPT-5.1 Thinking, giving users the ability to tweak tone and communication style. The update tackles earlier complaints about responses feeling too cookie-cutter and aims to make conversations flow more naturally.
U.S. data-center investment has surged more than 300% in three years, far outpacing all other commercial construction categories. New data shows AI-related spending contributed most of U.S. GDP growth in early 2025.
Oracle chairman Larry Ellison recently stated that today's leading AI models like ChatGPT, Gemini, and Grok run on public internet data but will only reach their full potential when trained on proprietary datasets. His remarks have put renewed focus on ORCL and other companies positioned to capitalize on private data ownership.
Fresh benchmark data reveals NVIDIA H100 and B200 delivering the lowest inference costs for Llama 3.3 70B, with Google TPU v6e and AMD MI300X falling behind in tokens-per-dollar efficiency.
Jeff Bezos' stealth AI initiative Project Prometheus has quietly acquired agentic computing startup General Agents after raising over $6 billion and expanding to more than 100 employees.
ChatGPT is gearing up for a massive search overhaul in 2026, with eight key algorithm changes that will transform how brands get discovered and ranked in AI-powered search results.
Anthropic's Claude Code version 2.0.55 improves fuzzy matching for file suggestions and fixes bugs in DNS resolution and keyboard navigation.
OpenAI released version 0.2.24 of the Agents SDK. The new version adds resumable streaming, better MCP APIs, quicker discovery and reliability fixes. It gives developers a steadier base when they run long tasks or use agents across multiple tabs.
Kimi's strong benchmark performance against major AI models has reignited discussions about its valuation, which sits 4-5 times below competing labs despite delivering competitive results.
Anthropic's latest research shows that its AI models deliberately cheat during safety tests. The evidence indicates that the quality of the training data causes this behavior, not reinforcement learning alone.
We use cookies to improve your experience on our site and to show you relevant advertising. To find our more, read our privacy policy and cookie policy
Eseandre Mordi
Saad Ullah
Sergey Diakov
Alex Dudov
Usman Salis
Peter Smith
