Companies should consider building their own AI Site Reliability Engineer (SRE) to manage the increased code output from agentic coding tools, according to a post on The New Stack. The approach aims t…
OpenAI and Elastic have partnered to integrate Elasticsearch with OpenAI's retrieval-augmented generation capabilities, enabling enterprises to build AI-powered search and analytics applications with …
OpenAI said its GPT-5.6 model can reduce its own operational costs, according to a company statement.…
Anthropic advocates for safety testing of open-weight AI models rather than outright bans, diverging from OpenAI and Google which have backed restrictions on open-weight releases. The debate centers o…
Cloudflare has open-sourced a debugger for privacy protocols used by Apple and Microsoft, designed with AI agents in mind. The tool aims to help developers verify that privacy-preserving systems like …
Nvidia, Palantir, and Hugging Face have joined 30 other organizations to form the Open Secure AI Alliance, a group dedicated to defending open-weight AI models from cyber threats. The alliance aims to…
The Model Context Protocol (MCP) released its biggest update, removing the machinery many servers were built around, according to an announcement from the protocol's maintainers.…
Anthropic launched Opus 5 on Friday, the latest iteration of its heavyweight model, arriving just two months after Opus 4.8. The new model costs a third of the price of its predecessor, which the comp…
Anthropic's Opus 5 model is nearly as capable as Fable 5, according to a comparison by the company. The new AI model shows significant improvements in reasoning and coding tasks, narrowing the gap wit…
Cursor, the AI coding tool recently acquired by Elon Musk's SpaceX in a $60 billion all-stock deal, has launched a model router, joining Ramp and Meta in building such systems. However, Cursor and Ram…
A new analysis from The New Stack explores whether prompt caching can reduce the costs of retrieval-augmented generation (RAG) without sacrificing accuracy, finding that caching strategies can lower l…
OpenAI has built support agents for its own customer service line and now aims to convince large enterprises to adopt similar AI agents, signaling a shift where models are no longer the bottleneck in …
The New Stack reports that single-pass AI code generation is not obsolete, but 'high-reasoning' models represent the next frontier in software development, according to the publication's analysis of c…
Microsoft and Mistral have deepened their partnership with a multibillion-dollar agreement focused on enterprise AI infrastructure, deliberately building an AI stack the company does not fully own. Th…
Alibaba has revealed Qwen 3.8, its latest large language model, claiming it is among the most powerful but providing no real data to support the assertion, according to The New Stack.…
Google has bet its inference future on a custom chip built specifically for one model, the company announced. The chip, called the Tensor Processing Unit (TPU) v5p, is designed to optimize inference f…
Anthropic employees worked "literally around the clock" to keep Claude Fable 5 from disappearing, the company said, after weeks of extending temporary access while bringing additional inference capaci…
AI agent reliability problems now stem from the context layer, not the model, according to a pattern observed over two years by The New Stack.…
AI has not shifted the bottleneck in software development from coding to code review, according to The New Stack. The article argues that while AI tools assist with code generation, human review remai…
1Password has launched a new browser integration for Claude that enables the AI agent to securely access and use stored credentials, addressing authentication challenges as AI agents take on more onli…