Liquid AI releases LFM2.5-230M small model for on-device inference

MarkTechPost

Liquid AI's smallest model at 230M parameters, available open-weight on Hugging Face with base and instruction-tuned checkpoints. Targets agentic tasks on phones, robots, and automation devices. Supports inference across llama.cpp, MLX, vLLM, SGLang, and ONNX.

lfm2-5LLaMA

MarkTechPost·Asif Razzaq

3 hours ago