Updated

LFM3

Liquid AI released LFM3, a new generation of open-weight Liquid Foundation Models that puts text, vision, audio, omni, and decision models on one shared backbone. The first release covers six 2B-class models trained from LFM3-1.7B-Base.

By No Code MBA · Facts checked

Provider
Liquid AI
Announcement date
Status at last check
Generally available
Model family
LFM3
Preceding model
LFM2.5
Capabilities
Text, Image, Audio

What changed

LFM3 uses a new tri-hybrid architecture and 22T tokens of pre-training. Liquid AI says LFM3-1.7B outperforms LFM2.5-2.6B with 35% fewer parameters, the text model fits in about 1 GB at 4-bit precision, and text context reaches 256K tokens.

Access and availability

Base and post-trained models are open-weight and available on Hugging Face, the Liquid Playground, and the Liquid AI API.

Official access documentation

What you can build

A reasonable starting point for teams that want a small model to fine-tune locally for one specialized task on device.

Limitations

The first LFM3 release covers only models in the 2B parameter range. Liquid AI positions them as starting points for specialization through fine-tuning rather than general frontier assistants.

Sources and release history