liquid/lfm-3-1.7b
A compact 1.7B-parameter text model with a 256K-token context, post-trained for agents, tool calling, and structured outputs.