← All models

LFM2-VL-1.6B

Liquid AI · 2025-08-12 · 1.6B parameters

LFM2-VL-1.6B is the larger of Liquid AI's first two vision-language models, released August 12, 2025: the LFM2-1.2B language model with a 400M-parameter shape-optimized SigLIP2 NaFlex vision encoder that reads images at native resolution. The card gives a 32,768-token text context and English support. The BF16 checkpoint contains 1,584,804,000 parameters, and the tokenizer assigns 64,400 distinct IDs within 65,536 embedding rows.

Benchmark scores