LangChain Tests NVIDIA Switchyard: 93% of Agent Calls Handled by 30B Model, Cutting Costs 70%

LangChain · x · 2026-08-12

LangChain integrated NVIDIA's open-source model router, Switchyard, into its Deep Agents evaluation suite, comprising 145 multi-step tasks involving tool use, retrieval, and long context.

Results: 93% of turns were successfully routed to a 30B model, while only 7% required Claude Opus 4.8. This routing strategy reduced total costs by 70% while retaining 90% of Opus's accuracy on the same tasks.

Related event: LangChain Tests Show Model Routing Cuts Agent Costs by Over 70%(2 posts)→

Original post →

More from coding & agent

coding & agent channel →