Liquid AI Releases LFM3: Six Open-Weight 2B Models
Liquid AI launches LFM3, featuring six 2B-class open-weight multimodal models supporting text, vision, audio, omni, and decisions with a 256K context window.
Stock photo for illustration only, not from the actual event
- Liquid AI releases LFM3 with six open-weight 2B-class models
- Supports text, vision, audio, omni, and decision-making tasks
- LFM3-1.7B utilizes a tri-hybrid backbone and a 256K context window
- Fits in approximately 1 GB of storage when quantized at 4-bit
Liquid AI has officially launched the LFM3 lineup, introducing six open-weight 2B-class models designed to handle text, vision, audio, omni, and decision-making capabilities. This new release aims to provide versatile solutions across various artificial intelligence modalities.
A standout model in this release is the LFM3-1.7B, which incorporates a tri-hybrid backbone architecture and supports an extensive 256K context window, enabling the processing of significantly larger inputs efficiently.
Stock photo for illustration only, not from the actual event
The introduction of Liquid AI's LFM3 series highlights the ongoing industry shift toward compact, highly capable multimodal small language models. By integrating a tri-hybrid architecture with a 256K context window in a 1.7B parameter size, developers can achieve broad contextual awareness while maintaining a lightweight footprint suitable for resource-constrained environments.
Regarding hardware efficiency, the LFM3-1.7B model requires only about 1 GB of storage space when operating at 4-bit quantization, lowering the barrier for deployment on edge devices and local servers.
Source: MarkTechPost
Found something wrong in this article? Report an issue with this article
Comments
Leave a Comment