What Inclusion AI does
Inclusion AI (inclusionAI) is Ant Group's open-source AI research lab. It develops and releases the Ling (also called BaiLing) model family, which spans three series: Ling non-thinking language models optimized for efficiency, Ring thinking models built for advanced reasoning, and the multimodal Ming series for understanding and generation. The lab has been shipping new open models at a rapid pace and distributes them publicly on Hugging Face and ModelScope.
Key capabilities
Inclusion AI's models use sparse mixture-of-experts and hybrid linear architectures to deliver strong performance with reduced token usage. Recent releases include Ling-2.5-1T, which supports context lengths up to 1 million tokens, and Ring-2.5-1T, a hybrid-linear model tuned for reasoning. The Ming series adds multimodal capability, with Ming-Flash-Omni-2.0 unifying speech, audio, and music in a single architecture. Models in the family are released under the permissive MIT license.
Who it's for
Inclusion AI serves researchers, developers, and enterprises that want frontier-scale open-weight models for reasoning, multilingual, and multimodal applications they can self-host or fine-tune.