importantSYS.SOURCE: Liquid AI• 2026-10-14T23:59:37Z
LFM2.5-24B-A2B: High-Performance On-Device Reasoning Model with 2B Active Parameters
Liquid AI releases LFM2.5-24B-A2B, a 24B-parameter on-device reasoning model with 2B active parameters, achieving competitive performance against larger models while offering 1.7× faster inference. The model is open-weight, available on Hugging Face, and demonstrates 184 tokens/s on Apple M5 Max and 21K output tokens/s on NVIDIA B300 GPU.
*** END OF TRANSMISSION ***