The development
What happened
Liquid AI announced LFM3, an open-weight family built around a shared 1.7-billion-parameter backbone. Its six roughly 2-billion-parameter variants target text, vision, audio, combined modalities and structured decisions. The reported specifications and resource requirements are vendor claims, not independently validated results.
Key facts
The six variants cover text, vision including video, audio, combined modalities and structured decisions.
Liquid AI says each variant can be fine-tuned on a single GPU and provides model cards, inference examples and fine-tuning notebooks.
The vendor says the text model supports a 256K context and occupies about 1 GB at four-bit precision.
The context
Why it matters
The family offers a common starting point for specialized multimodal applications, potentially with modest fine-tuning resources. Practical performance remains unverified by the supplied evidence, and open weights alone do not establish unrestricted licensing.