Ollama, the open-source tool for running large language models locally, released version v0.33.3-rc0 today. This release candidate focuses on internal updates and improved model handling. Key changes include an update to MLX, MLX-C, and llama.cpp components, which are crucial for model inference and performance. Additionally, Ollama v0.33.3-rc0 now honors GGUF model defined default parameters. This change improves consistency and accuracy when running models that specify their own defaults, ensuring they behave as intended by their creators. A new contributor, @marcelpetrick, made their first contribution to this release.