ownllm/phi-4-mlx-4bit

Phi-4 (14B), MLX 4-bit, for OwnLLM

A copy of the 4-bit MLX conversion of Microsoft's Phi-4 (about 14 billion parameters), kept in the OwnLLM model collection for Apple Silicon computers running MLX.

• Original model: microsoft/phi-4, by Microsoft, MIT license (see LICENSE).
• Conversion: MLX 4-bit, by the MLX community. The weights are unchanged by OwnLLM.
• Size: about 7.7 GB. Needs an Apple Silicon Mac with 16 GB of memory or more.
• Format: MLX. The OwnLLM package for Mac, Linux and Windows uses GGUF files with llama.cpp and does not load this format; these files are for Apple Silicon use with mlx-lm.

pip install mlx-lm
mlx_lm.generate --model ownllm/phi-4-mlx-4bit --prompt "Explain a mortgage in two sentences."

OwnLLM is not affiliated with or endorsed by Microsoft. A language model can be confident and mistaken; do not rely on it alone for medical, legal, financial or safety decisions. Project: https://github.com/ownllm-org/ownllm