Ollama v0.34.1 Released with MLX Memory Handling Improvements on Apple Silicon

Core summary

One sentence to understand this update

Ollama has released version v0.34.1, featuring improved MLX memory handling on Apple Silicon and updated GGUF model creation requirements to use llama.cpp tooling for safetensor conversion and quantization.

Impact & opportunity

What this could mean

Developers deploying local models with Ollama, especially on Apple Silicon devices, will experience improved performance and stability.