Llm
Framework-agnostic interface for streaming LLM inference.
Streaming counterpart of LlmMessageSender. Implementations handle the actual LLM communication (Spring AI, LangChain4j, etc.) and return a reactive stream of raw content chunks.
Key Differences from Non-Streaming:
Returns
Flux<String>instead ofLlmMessageResponseTool execution is managed by the underlying framework (e.g., Spring AI) since the streaming API is opaque - we cannot inject a custom ToolLoop
Only observation of tool execution is possible via ToolCallInspector
See also
for non-streaming equivalent
for tool execution observation