Tag
Ling-2.6-flash is a 104B-total/7.4B-active sparse instruct model optimized for token efficiency, aiming to cut costs and boost throughput on agent tasks.