Ollama released version 0.34.0 today, significantly enhancing its integration capabilities. The headline feature allows users to directly employ Ollama models within ChatGPT Desktop, facilitating the use of open models while preserving existing user workflows. The setup for this functionality is accessible via the Ollama application on MacOS. This release also delivers improved structured output performance for systems powered by Apple Silicon. Further additions include support for OpenAI-compatible client tool search and a new response compaction feature, which also correctly handles images through compacted responses.