Bug fixes
- Fixed: OpenAI-compatible local servers (LM Studio, LocalAI, text-generation-webui, etc.) were incorrectly treated as cloud backends, causing "API key not configured" errors
- Fixed: AI buttons blinking as unconfigured when only an OpenAI-compatible local server was set up
- Fixed: Default model picker in Local settings now shows both Ollama and OpenAI-compatible models
- Fixed: Cloud fallback cleanup when disabling the setting
- Fixed: Local OpenAI-compatible models no longer appear in the Cloud model picker