langchain-anthropic 1.4.8 fixes content block streaming bug

Full analysis

LangChain released version 1.4.8 of its Anthropic integration package, shipping a targeted bug fix: content_block_start events now correctly preserve initial text that was previously being dropped during streaming. This matters for agent builders using LangChain with Anthropic's Claude models, as dropped text at the start of a content block could silently corrupt streamed tool-call responses or reasoning output. The release also bumps the langgraph-checkpoint dependency to 4.1.1 and improves error messaging in LangChain core by adding message context to bare ValueError raises.