Ant Open Source announced, officially released and open-sourced Ling-3.0-Flash-VL, the first native multi-modal large model in the Bering series. The model was extended based on the LING-3.0-Flash MoE architecture. The total number of parameters is 124B, the single inference activates 5.5B parameters, natively supports image, text and video input, and the context window reaches 256 kToken.

Ant Open Source announced, officially released and open-sourced Ling-3.0-Flash-VL, the first native multi-modal large model in the Bering series. The model was extended based on the LING-3.0-Flash MoE architecture. The total number of parameters is 124B, the single inference activates 5.5B parameters, natively supports image, text and video input, and the context window reaches 256 kToken.