Skip to content
Trending storyDeveloping

LangChain's Open SWE routes tasks to cheapest passing model, cutting median cost 64%

1 article1 sourcesince Oct 10Last article 1h ago ·

Overview

AISummary of 1 article

Harrison Chase says LangChain's Open SWE now moves model choice into the harness, sending each task to the cheapest model that passes quality tests.

He reports median cost per task dropped 64%.

In a quoted reply, Chase says DeepSeek v4.1 Flash handles orchestration, implementation, and verification well, with frontier models kept for harder edge cases.

Written by AI from the articles below · updated Oct 10, 12:41 AM ET

Check the sources:

Latest developmentOct 10, 12:31 AM ET
LangChain's Open SWE routes each task to the cheapest adequate model

Article timeline

The articles in this story. Times are ET.

Oct 10
  1. Harrison ChaseX
    LangChain's Open SWE routes each task to the cheapest adequate model

    AILangChain says Open SWE now moves model choice into the harness, sending each task to the cheapest model that passes quality tests. Median cost per task dropped 64%. The post's quoted reply says DeepSeek v4.1 Flash handles orchestration, implementation, and verification well, with frontier models kept for harder edge cases.

Heat trend

Not enough continuous observations to show a trend yet.