v0.38.3

Release v0.38.3

Release Notes: Hexai v0.38.3## OverviewThis is a targeted patch release focusing on optimizing the default experience for users utilizing the Ollama provider. The primary goal of this update is to ensure users are routed to the most capable default model available in the cloud.## Improvements### LLM Provider Defaults*Updated Ollama Default Model: The default model for the Ollama provider has been updated from gemma4 to gemma4:31b-cloud.*Why this is useful: This change ensures that users who haven't specified a custom model benefit from the improved reasoning and performance of the31B cloud-hosted version of Gemma4, providing higher quality completions and more accurate code actions out of the box.## Technical DetailsThe changes are confined to the internal LLM client utilities and the Ollama provider implementation to ensure consistent fallback behavior across the CLI and LSP server.Modified files:- internal/llm/ollama.go: Updated hardcoded fallback model.- internal/llmutils/client.go: Updated provider default mapping.- internal/llmutils/client_test.go: Updated test assertions to match the new default.- internal/version.go: Version bump to 0.38.3.## Migration & CompatibilityThere are no breaking changes in this release. Users who prefer the smaller gemma4 model or a different specific version can continue to override the default in their config.toml file:toml# Example override in config.tomlOllamaModel = "gemma4"