TL;DR: Ollama's latest release, v0.33.3, integrates updates to MLX, MLX-C, and llama.cpp, enhancing local model serving capabilities.
Summary: Ollama has released version 0.33.3, which includes significant updates to its underlying MLX, MLX-C, and llama.cpp components. This release also adds features like reporting cached prompt tokens and honoring GGUF model-defined default parameters.
Why it matters: AI builders using Ollama for local model deployment will benefit from improved performance and compatibility due to these core library updates. Experiment with the new version to leverage potential optimizations and better control over GGUF model parameters.
Source: github_releases