Message content is a str or list of blocks: use .text, content_blocks wraps unknown block types as non_standard
AIMessage(content=[{'type':'text','text':'hello '},{'type':'text','text':'world'}]).text == 'hello world' while str(content) returns the list repr; .text skips non-text blocks (a 'thinking' block plus a text block gives only the text). AIMessage(content='hi').content_blocks == [{'type':'text','text':'hi'}], and a legacy {'type':'thinking',...} block appears in content_blocks as {'type':'non_standard','value':{...}}. HumanMessage(content=123) raises a ValidationError (2 validation errors: content.str and content.list). Chunk addition: AIMessageChunk('a')+AIMessageChunk('b') gives 'ab'; list blocks with the same index merge text ('ab'); str plus list gives a mixed list ['a', {...}]. Tested offline on Windows 11, Python 3.12.10, langchain-core 1.8.0, langchain 1.5.0, langsmith 0.14.17, langchain-google-genai 4.5.0, google-genai 3.3.0, pydantic 2.14.3.