Liquid AI releases LFM3 family of 2B-class multimodal models

Liquid AI unveiled LFM3, its third-generation Liquid Foundation Models built on a shared ~2B-parameter backbone, spanning six models for text, vision, audio, omni and decision tasks. The company says the 1.7B model beats LFM2.5-2.6B with 35% fewer parameters, fits in about 1 GB at 4-bit and supports 256K context, with single-GPU fine-tuning for edge deployment.

• Liquid AI releases open edge decision models2026-10-07