LING-3.1-FLASH (HUGGING FACE REPO)

Summary from the TLDR AI newsletter

Ling-3.1-flash is a native hybrid reasoning model designed for complex, real-world tasks with 560 billion total parameters, 25 billion active parameters per token, and a one-million-token context window.