1662 articles
Nano Banana 2 has unexpectedly shown up in Preview mode on Media IO. Multiple users have confirmed access, though experts warn it might not stick around for long.
ChatGPT maintains its position as the world's fifth most-visited website in October 2025, outpacing major platforms like X and Reddit while demonstrating consistent monthly growth.
A new analysis suggests global token generation is accelerating so fast that machine-created text could surpass all human speech by 2032—raising questions about how governments might regulate or tax an economy built on exponentially expanding AI data.
A new image-generation model from Nano Banana/GemPix 2 is turning heads after nailing two notoriously tough visual accuracy tests in a single image—something most AI systems struggle to do.
Google has launched DS-STAR, an advanced AI agent that automates data science workflows and handles messy, multi-file datasets with unprecedented accuracy, achieving top benchmarks in real-world data analysis tasks.
A new AI model believed to be GPT-5.1 takes the lead in creative writing as crypto mining faces regulatory challenges.
Fresh code snippets from OpenAI's Enterprise RBAC system hint that three GPT-5.1 models are on the way. Developer screenshots show internal descriptions for GPT-5.1, GPT-5.1 Reasoning, and GPT-5.1 Pro.
Anthropic's revenue has skyrocketed over ninefold in just one year, prompting the AI company to expand its workforce with new local teams to keep up with surging demand.
The Magnificent 7 now control a record 35.9% of the S&P 500's market cap, pushed higher by AI excitement, while generating just 26.8% of actual profits.
Morgan Stanley predicts private credit will supply over half of the $1.5 trillion needed for data center expansion, fundamentally changing how AI infrastructure gets funded.
Kimi K2 Thinking has delivered industry-leading performance across major AI benchmarks, surpassing GPT-5 and Claude Sonnet 4.5 in several categories while offering dramatically lower cost.
Anthropic is expanding into Paris and Munich as its EMEA revenue jumps over ninefold in just one year. The company's hiring local teams to meet the region's surging demand.
ERNIE 5.0 Preview has grabbed the No. 2 spot on the LMArena Text Arena leaderboard with a score of 1432, marking a significant achievement for Baidu's next-generation AI model.
LlamaIndex and MongoDB unveiled a workflow showing how LlamaParse transforms messy enterprise documents into clean, structured data for LLM applications.
Anthropic is making waves by posting a Head of Product position with a $600–650K base salary directly on LinkedIn instead of going through traditional executive recruiters. This move reflects a shift in how senior roles are filled in the AI industry.
Moonshot AI's Kimi K2 Thinking model is being hailed as the best open-source system after breaking multiple benchmark records, standing in stark contrast to OpenAI's expensive proprietary approach.
Google launched File Search, a managed RAG system in the Gemini API that delivers fast document indexing, vector search, and cited answers. It handles PDFs, code files, and structured data with easy setup.
Perplexity has upgraded its Comet Assistant with better performance, smarter workflow handling, and safer browser automation. The update lets users tackle longer, more complex tasks across multiple tabs with a 23% performance improvement.
Google DeepMind has rolled out a fully managed Retrieval-Augmented Generation (RAG) system within the Gemini API, making it significantly cheaper and easier to build RAG workflows. Developers can now upload files, search through them, and get cited responses without dealing with infrastructure headaches.
Microsoft's new study reveals which 40 jobs face the highest AI disruption risk and which remain safe, showing writing, translation, and analytical roles under greatest pressure.
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