LangSmith's Latest Feature: Grouped Monitoring Charts

LangSmith has introduced a new feature, grouped monitoring charts, to compare performance and feedback metrics for LLM applications. This allows users to group logged traces by tags and metadata to analyze different versions of their applications side-by-side. A case study on the Chat LangChain bot demonstrates how this feature can compare latency and time-to-first-token across different LLM providers like Claude, Mixtral, and Gemini. This capability also supports other use cases such as A/B testing revisions or analyzing user-specific experiences by grouping data by user or conversation IDs.