How to use from
llama.cpp
Install (macOS, Linux)
curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf manateelazycat/MiniCPM5-2B-AIPod:Q8_0
# Run inference directly in the terminal:
llama cli -hf manateelazycat/MiniCPM5-2B-AIPod:Q8_0
Install from WinGet (Windows)
winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf manateelazycat/MiniCPM5-2B-AIPod:Q8_0
# Run inference directly in the terminal:
llama cli -hf manateelazycat/MiniCPM5-2B-AIPod:Q8_0
Build from source code
git clone https://github.com/ggml-org/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-app
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama serve -hf manateelazycat/MiniCPM5-2B-AIPod:Q8_0
# Run inference directly in the terminal:
./build/bin/llama cli -hf manateelazycat/MiniCPM5-2B-AIPod:Q8_0
Use Docker
docker model run hf.co/manateelazycat/MiniCPM5-2B-AIPod:Q8_0
Quick Links

MiniCPM5-2B for AI Pod

Official MiniCPM5-2B Q8_0 weights (Apache-2.0), accepted AGX Thor / T5000 and AGX Orin Docker-save runtime archives, complete control-plane delivery, and a minimal model LPK. Each third-party component retains its included license. Original model: https://huggingface.co/openbmb/MiniCPM5-2B-GGUF Do not install a runtime archive as model weights. The LPK selects one platform for both weights and the hardware-specific runtime; all downloads are pinned by commit, size and SHA256. Inference runs offline with a read-only model mount. This repository is mirrored byte-for-byte on Hugging Face and ModelScope.

Install and inspect this release

Install the LPK in LazyCat and select AGX Thor / T5000 or AGX Orin in AI Pod. The bootstrap verifies its external control plane before starting; Host verifies weights and the matching runtime from the selected provider. Model-specific runtime content is hosted entirely on Hugging Face and ModelScope.

Downloads last month
58
GGUF
Model size
3B params
Architecture
llama
Hardware compatibility
Log In to add your hardware

8-bit

Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support