AI Intel

An arXiv paper proposes an adaptive, tri-metric routing framework that moves beyond static RAG to enable efficient long-context inference on commodit…

The Servo project blog reflected on one year of sponsored development.

An update with no specific title or content was noted on Hacker News.

Manticore Search has introduced internal chunking to improve the efficiency of vector search for long documents.

ChatGPT Work has introduced a new Data agent, enabling users to transform company data into answers, interactive dashboards, and actionable insights…

OpenAI has released GPT-Live-1 into its API, allowing developers to integrate ChatGPT's natural, conversational voice capabilities into their applica…

Ollama v0.34.1 is out, bringing MLX safetensors support, making `ollama create` stable, and improving MLX memory handling for Apple Silicon.

Gemini Code Assist announces General Availability for Gemini 2.5 Pro and Flash models, alongside new features like agent mode for VS Code, chat code…

GoogleAI has launched Gemini 3.8 Flash, described as their most intelligent workhorse model, offering significant upgrades in coding, agentic workflo…

Kimi_Moonshot's open-weight model, Kimi K3, is now generally available and integrating into GitHub Copilot, demonstrating frontier-level agentic codi…