Langfuse released v4 GA across cloud and self-hosted deployments, featuring a rebuilt foundational data model, a high-density observations table, full-text search, code-based evaluators, and automated production monitors.

Key Takeaways

  • ✓Foundational schema rewrite significantly accelerates high-throughput trace ingestion and multi-filter querying;
  • ✓Code Evaluators enable deterministic Python/TypeScript assertion logic to test agent outputs rigorously;
  • ✓Transitions Langfuse from passive trace logging into an active production quality gate and SLO monitoring platform.