v1.6.1

Release Notes - Version 1.6.1

Features

• Text response streaming: Added streaming support for text responses, enabling real-time token delivery for improved user experience

Bug Fixes

• Enhanced streaming compatibility: Fixed streaming failures with strict OpenAI-compatible servers (including older Ollama builds, llama.cpp, LocalAI, and some proxies) that previously rejected unknown request fields. The system now automatically retries without incompatible options when servers return errors, ensuring streaming works across a wider range of providers

Improvements

• Improved handling of different chunk formats from various providers including Anthropic-compatible servers, Ollama, and vLLM
• Added automatic fallback mechanism for servers that don't support advanced streaming options