Liquid AI Releases LFM2.5-DSpark Draft Models That Deliver Up to 3.18x Faster Decoding Without Changing Model Outputs

Liquid AI has introduced LFM2.5-DSpark, a set of three draft models designed to accelerate the performance of LFM2.5. By utilizing speculative decoding, these 300M parameter models allow the system to generate text up to 3.18 times faster while maintaining the same output as the original model.