Floaty 0.0.15

Release Notes - Version 0.0.15 🌟

Enhancements

• Multi-Provider Support:

  • Introduced the ability to configure multiple AI providers, each with their own settings tab in the UI. Providers include OpenAI, Anthropic, Gemini, Azure OpenAI, and several others.
  • Configurations allow assignments for chat, embeddings, and screenshot captioning to be managed independently.
  • Facilitates the use of local models for high-frequency tasks, while cloud models can be used for chats.

• Provider Architecture Overhaul:

  • AiClientFactory: Centralized model client construction, significantly improving modularity and flexibility.
  • ProviderPresets: Pre-configured provider settings available out-of-the-box for quick setup and deployment.

New Local Model Capabilities

• Local Models Integration:

  • Implemented OnnxEmbeddingGenerator to run sentence-transformer ONNX exports locally.
  • New local model options shipped include BGE Small/Base and MiniLM L6.
  • Enhanced download service for model retrieval with progress tracking.

• Local Memory Management:

  • The system no longer requires an API key for core features like memory and screen history services, which now intelligently determine capacity based on configuration.

Miscellaneous

• README and documentation updates to reflect architectural changes and new feature capabilities. 📖
• Added new dependencies:
  • Anthropic C# SDK, Azure.AI.OpenAI, Microsoft.ML.Tokenizers, enhancing AI provider integration.

Collaboration

• Co-authored improvements by Claude Opus 5 with contributions from Joachim Leonfellner and nor0x.

Repository

• Explore more and contribute on GitHub: Floaty Repository

For more detailed information on the changes, please refer to the updated project documentation and code repository. Thank you to all contributors who made this release possible! 🚀