1662 articles
Third-quarter earnings filings indicate that Amazon and Google together hold nearly 17% of Anthropic, exposing the depth of Big Tech's financial commitment to one of the fastest-growing AI companies globally.
Chinese tech giant Meituan has released LongCat-Flash-Omni, a powerful open-source multimodal AI model that rivals Google's Gemini 2.5 Pro. With a 128K context window and real-time audio/video capabilities, it signals China's growing presence at the forefront of global AI innovation.
Epoch AI discovered a bug in its benchmarking code that was incorrectly labeling GPT-5's "high reasoning" mode as "medium." After the fix, both GPT-5 variants now score identically on the Epoch Capabilities Index—raising questions about how we measure AI performance and what "reasoning levels" actually mean.
Cognition and Windsurf's new SWE-1.5 model is turning heads with its blazing 950 tokens per second inference speed and agent-native design. Built on GLM-4.6 and deployed on cutting-edge Cerebras hardware, it could redefine what real-time AI coding assistance looks like.
New research from Epoch AI reveals that open-weight AI models are now reaching state-of-the-art performance just 3.5 months after their closed-source counterparts—a dramatic acceleration that's reshaping the competitive landscape of artificial intelligence.
OpenAI's Codex CLI v0.53 adds an experimental Windows sandbox for safer agent execution with better security and less manual oversight.
A look at how Meta, Google, Amazon, and Microsoft are reshaping the tech landscape with unprecedented infrastructure investments — and what it might mean for everyone else.
A major new study shows that generative AI increases online retail sales by up to 16.3% without requiring additional resources — adding roughly $5 in value per customer annually and providing solid proof of real productivity gains.
Chinese AI startup MiniMax has launched M2, claiming it's the first real "agent-native" AI model. Early reports suggest it outperforms Claude 4.1, Gemini 2.5, and Qwen3—while costing just 8% of Claude's price. If true, this could signal a major shift from chatbots to autonomous AI agents.
Nvidia is supplying over 260,000 AI chips to Samsung, Hyundai, and SK Group through a government-backed South Korean initiative to strengthen national AI infrastructure.
OpenAI releases Codex-CLI 0.52 with enhanced terminal integration, improved image processing, and inline command execution capabilities for AI-assisted development workflows.
Recent breakthroughs NVIDIA demonstrate that distributed AI training across continental distances is now feasible, potentially reshaping the future of artificial intelligence infrastructure.
Oracle's debt-to-equity ratio has hit nearly 500%—far beyond AI competitors—as the company lines up a $38 billion financing deal to build new data centers for AI infrastructure.
Google has upgraded NotebookLM with a 1 million-token context window, 6× longer conversation memory, and goal-based controls that transform it into an intelligent research companion.
Questflow has unveiled Daydreams, a groundbreaking platform where AI agents can operate independently—thinking, acting, and making payments without human input. Powered by x402 payment rails and connected to leading AI models, this marks a major step toward truly autonomous digital economies.
A silent Google Search update in September 2025 removed a URL parameter that let users view 100 results per page, replacing it with infinite scroll. The change slashed Reddit's valuation by an estimated $3 billion and fundamentally altered how AI models access web content.
Anthropic has released a native installer for Claude Code that eliminates the need for Node.js, making it faster and easier for developers to get started with the AI coding assistant.
Microsoft CEO Satya Nadella announced a major upgrade to Researcher, introducing "Computer Use" — a feature that lets the AI assistant securely browse both open and gated websites, handle multi-step tasks, and generate deeper reports.
Researchers have unveiled DeepAgent—a breakthrough reasoning model that can think, discover tools, and act on its own—outperforming GPT-4-level agents across major AI benchmarks.
MiniMax M2 paired with Claude Code has shot up to second place on the AI CodeKing leaderboard, beating out GLM-4.6 and other coding agents in both performance and cost efficiency.
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
Usman Salis
Peter Smith
Sergey Diakov
Alex Dudov
