Today · October 10

LangChain cut Open SWE's median cost 64%; it's self-reported

LangChain CEO Harrison Chase says moving model choice into the harness of Open SWE, the company's open-source coding agent (an AI that takes actions itself), so each task goes to the cheapest model that passes a quality bar, cut median cost per task by 64%. He writes: "most orchestration steps don't need a frontier model. for Open SWE we moved model choice into the harness, so each task goes to the cheapest model that still does the job, tested against quality. median cost per task dropped 64%" (Harrison Chase, 2026-10-10). The harness is the layer of code wrapped around the model that manages prompts, tools and flow. LangChain's October 1 blog post sets out the comparison: the control "always used GPT-6 Astra"; quality was measured by whether a thread ended in a merged pull request, 29.2% for routed threads against 27.3% for the control (p = 0.49, not a significant difference); and the median routed thread cost US$0.94 against US$2.61 for the control, "64% less" (LangChain blog, 2026-10-01). This is still one product's self-reported median and can't be extrapolated to other agents. Teams that want to try the same thing can first measure how often cheaper models pass on their own tasks, then decide which steps can be downgraded.

Concept illustration: an open laptop sits on a light desk at left, its screen showing a speech bubble with grey lines and its palm rest lettered OPEN SWE; a red cable runs from the laptop's side down to a small black box with a port and three dots, which stands next to a tall dark rack of eight stacked units on the right. A pale banner near the top shows a green parrot, a green chain-link icon and the lettering LANGCHAIN.

Source

This item ran in the brief of Oct 11, 2026, alongside 13 other items.

Cite this item

SecondSource (2026). "LangChain cut Open SWE's median cost 64%; it's self-reported." https://secondsource.io/en/product/2026-10-11-1/. Accessed access date.