TL;DR: Ollama's latest release enhances Gemma 4 tool calling and multi-turn reasoning, making local AI agents more capable.
Summary: Ollama v0.32.1 significantly improves Gemma 4 tool calling and multi-turn reasoning, including more reliable tool-response continuations. This update also addresses an MLX model cache leak, improving memory management for local AI model inference.
Why it matters: AI developers using Ollama can now build more robust and intelligent agents with Gemma 4, especially for complex, multi-step tasks. Experiment with the improved tool calling to create more sophisticated local AI applications.
Source: github_releases