Latest Blog

All
Official ggml-org llama.cpp v0.8.0 GitHub release card highlighting LFM3 and experimental streaming audio
llama.cpp 0.8.0: New Models, Real-Time Voice, and Local AI Features

llama.cpp 0.8.0 adds LFM3 and other model architectures, experimental real-time audio APIs, TTS support, and server and memory improvements. This guide separates the verified release changes from deployment limits, then shows a reproducible local validation workflow.

Read more
Official Liquid AI LFM3 launch artwork: Every Modality for Specialized Intelligence
Liquid AI LFM3 Explained: Can a 2B Multimodal Model Run Locally?

Liquid AI’s LFM3 is a family of compact open-weight text, vision, audio and omni models. Learn what the 1 GB 4-bit text footprint really means, how to size RAM and run LFM3 locally with llama.cpp, and...

Read more