llm-gemini 0.34: Gemini 3.8 Flash model and fix for asynchronous version recording

The llm-gemini developers released version 0.34, which includes a new model and bug fixes. The release centers on the llm-gemini project and the new gemini-3.8-flash model, which brings the Gemini 3.8 Flash variant. The new model arrives with three reasoning levels—low, medium and high—providing more flexible configuration for different tasks. The release references issues #146 and #137, one of which fixes a problem in asynchronous responses where the resolved model version was not recorded; the fix was described with the involvement of Charlie Tonneslan. The explicit version and model names (0.34, gemini-3.8-flash and Gemini 3.8 Flash) help identify the release and compatibility with user applications. The changes are important for reproducibility and version tracking, especially for asynchronous processing. The new reasoning levels may enable cost and performance optimization per application. Overall, the update expands modeling options and improves reliability in version management.