The agent engineering platform.
Python agentsaiai-agentsanthropicchatgptdeepagentsenterpriseframeworkgeminigenerative-ailangchainlanggraphllmmultiagentopen-sourceopenaipydanticpythonragtypescript
revert: use usage metadata scaling in SummarizationMiddleware default token counter (#35002)
ccurme643355fa2df2
diff --git a/libs/langchain_v1/langchain/agents/middleware/summarization.py b/libs/langchain_v1/langchain/agents/middleware/summarization.pyindex ab36cd554c..8e9edca3e6 100644--- a/libs/langchain_v1/langchain/agents/middleware/summarization.py+++ b/libs/langchain_v1/langchain/agents/middleware/summarization.py@@ -144,10 +144,8 @@ def _get_approximate_token_counter(model: BaseChatModel) -> TokenCounter:if model._llm_type == "anthropic-chat": # noqa: SLF001# 3.3 was estimated in an offline experiment, comparing with Claude's token-counting# API: https://platform.claude.com/docs/en/build-with-claude/token-counting- return partial(- count_tokens_approximately, chars_per_token=3.3, use_usage_metadata_scaling=True- )- return partial(count_tokens_approximately, use_usage_metadata_scaling=True)+ return partial(count_tokens_approximately, chars_per_token=3.3)+ return count_tokens_approximatelyclass SummarizationMiddleware(AgentMiddleware):