Ollama v0.30.6 integrates Oh My Pi and improves Apple Silicon quantization

Ollama, the popular open-source tool for running large language models locally, released version 0.30.6 with two notable enhancements. First, the `ollama launch omp` command now integrates with Oh My Pi, an AI coding agent that provides IDE integration. This allows developers to leverage local LLMs for code completion, refactoring, and debugging directly within their editor environment, without sending data to external APIs.

Second, the MLX (Apple's machine learning framework for Silicon) embedding layers now utilize NVFP4 global scale, a quantization method that improves memory efficiency and performance on Apple Silicon Macs. This means running embedding models locally becomes faster and uses less RAM, particularly beneficial for RAG pipelines and semantic search tasks. The update also includes various bug fixes and performance improvements from the v0.30.5 baseline. With over 173,000 GitHub stars, Ollama continues to be a key player in democratizing local AI, and this release reinforces its commitment to developer experience and platform optimization.