Source record

Published by LangChain Blog (T1.5 source). The original is at https://www.langchain.com/blog/building-long-running-agents-with-scaledown-and-deep-agents.

Pipeline notes

The summary and note below are generated by the signal pipeline — they are Beyond Desk’s reading, not quotations from the source.

ScaleDown's constrained decoding models now integrate with LangChain and Deep Agents via three middlewares: one summarizes old history, one compresses large tool results, and one extracts structured data. ScaleDown claims its models are 200x faster and 467x cheaper than frontier models on these narrow context-management tasks.

Concrete middleware pattern for context management with specific speed and cost numbers; relevant if long-running agents hit context or budget limits.

Cited by

No citations on record.