On device AI FTW!
Introducing 饾櫥饾櫟饾櫦饾煿: our new generation of Liquid Foundation Models. With a new tri-hybrid architecture and new multimodal encoders, the LFM3 family is the ideal model suite for specialized intelligence.View quoted post on X
> 6 model instances, across text, audio, vision, and omni modalities, generative and decision models.
> 饾櫥饾櫟饾櫦饾煿-饾煼.饾熃饾櫛: text, 256k context window.
> 饾櫥饾櫟饾櫦饾煿-饾殔饾櫥-饾煾饾櫛: vision-language model with video-input support for spatial understanding
> 饾櫥饾櫟饾櫦饾煿-饾櫚饾殲饾殟饾殥饾殬-饾煾饾櫛: tool calls, voice chat, reasoning in 18 languages.
> 饾櫥饾櫟饾櫦饾煿-饾櫨饾殩饾殫饾殥-饾煾饾櫛: an omni model combines vision, audio, and language.
> 饾殟饾煼-饾煾饾櫛 and 饾殟饾煼-饾櫨饾殩饾殫饾殥-饾煾饾櫛: decision models with typed outputs and calibrated probabilities in a single pass. Omni adds audio input.
> fine-tune on 1 GPU.
> a new tri-hybrid architecture
> architecture optimized for 14 processor classes across GPUs, CPUs, and NPUs.
> 饾櫥饾櫟饾櫦饾煿-饾殔饾櫥-饾煾饾櫛 is a beast. It performs competitively with Gemma 4 12B (6x larger), and Gemini 3.5 Flash Lite, and GPT-5.6 Luna, with 1.9B parameters.
> 饾櫥饾櫟饾櫦饾煿-饾煼.饾熃饾櫛 is #1 on the Pareto frontier of device intelligence 饾殭饾殥饾殭饾殠饾殱饾殱饾殠.饾殨饾殥饾殮饾殲饾殥饾殟.饾殜饾殥
Every modality for specialized intelligence. Pick the one that fits your use case, customize it, and run it on-device.
1/ 馃У