MiniCPM5-1B - A new SOTA for compact open models on the edge
MiniCPM5-1B is a dense 1B open model built for on-device and local deployment. It supports 131K context, Think / No Think modes, tool calling, GGUF and MLX formats, major inference backends, and even powers an offline desktop pet.
Replies
Flowtica Scribe
Hi everyone!
MiniCPM5-1B is currently the strongest open-source model under 2B for on-device use:
It hits SOTA in the 1B-class on agentic tool use, code generation, and tough reasoning tasks while keeping a very small footprint.
The INT4 weights are only around 0.5GB, which makes the local story much more real.
OpenBMB also shipped a cute Desktop Pet fully powered by this model — completely local, no cloud:
SOTA at 1B parameters running fully on device is wild. the cost of not needing cloud inference adds up fast when you're running agents all day. 131K context on edge hardware is the part I'd want to stress test