Ling-3.0-flash has 124B total parameters, while activating only 5.1B per token.
This work extends the same efficiency principle beyond model architecture: kernels, quantization, decoding, and hardware must work together to turn more of every unit of compute into useful intelligence.
Ant Ling releases Ling-3.0-flash with 124B total and 5.1B active parameters
AIAnt Ling says Ling-3.0-flash has 124B total parameters but activates only 5.1B per token. The company says the efficiency gains come from co-designing kernels, quantization, decoding, and hardware, not model architecture alone.
@AntLingAGI
Part of a thread · earlier post
Source details
Ant Ling · x.comPublished · added here
Original title: Ling-3.0-flash has 124B total parameters, while activating only 5.1B per token.