Liquid AI's new LFM2.5-Encoder runs 3.7x faster than ModernBERT on CPU at 8K-token context.
Continue to AI University →