TL;DR: Ollama v0.40.4 integrates the latest llama.cpp, adding support for decision models and EmbeddingGemma 2, enhancing local LLM capabilities.
Summary: Ollama has released version 0.40.4, updating its llama.cpp backend to b11521. This update specifically introduces support for 'decision models' and EmbeddingGemma 2. The release also includes improvements for reusing verified manifests and loaded models in inference routes.
Why it matters: This update expands the range of models and functionalities available for local inference, particularly for those working with Gemma embeddings and specialized decision-making models. Developers should explore how these new capabilities can be leveraged in their local AI applications.
Source: github_releases