Ollama v0.20.6 Fixes Gemma 4 Tool Calling and Adds Hermes Agent Integration Guide | Tool Update
Ollama v0.20.6 ships improved Gemma 4 tool calling using Google's post launch fixes, better parallel tool calling for streaming responses, and a new official H…
Published on MyPrivateClaw
Apr 13, 2026, 7:02 PM UTC
Coverage date
Apr 12, 2026
Last updated
Apr 13, 2026, 7:02 PM UTC
News summary
Ollama v0.20.6 was published on April 12, 2026. The release focuses on two areas: Gemma 4 reliability and agent integration documentation. Gemma 4 Tool Calling The previous v0.20.4 and v0.20.5 releases enabled flash attention for Gemma 4 on compatible GPUs. v0.20.6 goes further by applying Google's post launch fixes to improve tool calling accuracy and fixing parallel tool calling for streaming responses — an issue that caused dropped or duplicated tool calls when multiple tools were invoked in a single turn. Hermes Agent Integration Guide A new $1 is now live in the Ollama docs. This is the first official documentation for running Hermes Agent with Ollama as the local inference backend, covering setup, model selection, and the ollama launch workflow introduced in v0.20.5. App Fix The Ollama desktop app received a fix for image attachment errors that affected multimodal workflows on mac…