https://news.kuzmin.nyc/en/Fresh AI research and news: the original, the essence, and a beginner's explanationen https://news.kuzmin.nyc/en/item/2026-10-25-opends-world-deep-search/ https://news.kuzmin.nyc/en/item/2026-10-25-opends-world-deep-search/ Sun, 25 Oct 2026 12:00:00 GMT Paper The new OpenDS-World benchmark tests whether an agent can find and connect evidence across videos, PDFs, tables and websites on its own. Even the best model fails 122 of 200 tasks. https://news.kuzmin.nyc/en/item/2026-10-25-naive-rag-vs-agentic-memory/ https://news.kuzmin.nyc/en/item/2026-10-25-naive-rag-vs-agentic-memory/ Sun, 25 Oct 2026 12:00:00 GMT Paper Plain retrieval with reranking over raw conversation history beat twelve specialised memory systems in six of seven settings and runs many times faster. https://news.kuzmin.nyc/en/item/2026-10-25-kimi-k3-1-pro/ https://news.kuzmin.nyc/en/item/2026-10-25-kimi-k3-1-pro/ Sun, 25 Oct 2026 12:00:00 GMT News Moonshot AI's new model has a 1M-token context and, by the company's own measurements, nearly matches Claude Opus 5.5 on terminal coding tasks. The weights are not public yet.