# Ling-2.6-flash

Efficiency Over Scale: Bailing Unveils Ling-2.6-flash to Disrupt the Intelligence-Cost Curve

Bailing has launched Ling-2.6-flash, a 104B parameter model that uses Mixture of Experts (MoE) technology to activate only 7.4B parameters. It achieves benchmark parity with larger models while consuming only 10% of the tokens required by competitors like Nemotron-3-Super.