A KV cache, or key-value cache, speeds up autoregressive transformer inference by storing attention information calculated for tokens that have already been processed. Without the cache, the model would recompute those keys and values each time it generates another token. Reusing them reduces repeated work and improves response latency, especially for long prompts and outputs. The tradeoff is memory: cache size grows with sequence length, batch size, layer count, and model dimensions. Production systems manage this cost with paging, quantization, eviction, prompt caching, and request scheduling. KV caching changes inference efficiency, not the model’s learned knowledge, and cached data must be protected when prompts contain sensitive information.
Elon Musk has unveiled Grokipedia – an AI-driven encyclopedia built on Grok that aims to offer factual, bias-free knowledge as an alternative to Wikipedia.
A study by the University of Maryland and Microsoft found that Polish outperforms English and other major languages in long-context AI prompts, surprising researchers.
OpenAI has unveiled ChatGPT Atlas, a new web browser with ChatGPT integrated directly into its interface, blending AI assistance, search, and memory into everyday browsing.
Neobank Revolut has acquired Berlin-based AI travel agent start-up Swifty, integrating its conversational booking technology into Revolut’s loyalty and lifestyle platform for over 65 million users.
Andrew Tulloch, co-founder of the $12 billion AI startup Thinking Machines Lab, has joined Meta after previously rejecting what reports described as a $1.5 billion offer — a figure Meta has since called ‘inaccurate and ridiculous.’
Oracle has introduced an AI Agent Marketplace within its Fusion Applications suite, expanding AI Agent Studio and integrating top LLMs from OpenAI, Anthropic, and others to accelerate enterprise AI transformation.
AI presentation startup Prezent has raised $30 million to acquire services companies and integrate domain expertise into its enterprise communication platform, beginning with Prezentium.
Oracle Cloud Infrastructure will deploy 50,000 AMD Instinct MI450 GPUs beginning in 2026, underscoring the growing competition to Nvidia in the global race for AI computing capacity.
OpenAI has announced a strategic partnership with Broadcom to design and deploy 10 gigawatts of custom AI accelerators, expanding its hardware ecosystem beyond Nvidia and AMD.
Google has unveiled Gemini Enterprise, a next-generation AI platform designed for business productivity and collaboration, positioning itself directly against Microsoft’s Copilot and OpenAI’s enterprise solutions in the race for workplace AI dominance.