inclusionAI ships Ling-3.0-tiny, an 8B MoE model

inclusionAI open-weighted Ling-3.0-tiny on Hugging Face — an 8B-parameter MoE with 1.3B active, a smaller sibling of Ling-3.0-flash released days earlier. Reddit testers place it between 4B and 8-12B Qwen/Gemma models in performance, with high tokens/sec.