MLflow 3.7.0rc0 includes several major features and improvements!

Major Features

• ⚖️ Trace Comparison: New UI feature allowing side-by-side comparison of traces to analyze and debug LLM application behavior across different runs. (#17138, @joelrobin18, @daniellok-db)
• 💬 Multi-turn conversation support for Evaluation: Enhanced evaluation support for multi-turn conversations in mlflow.genai.evaluate, enabling comprehensive assessment of conversational AI applications. (#18971, #19039, @AveshCSingh)
• 🔎 Full Text Trace Search from UI: Search across all trace content directly from the UI, making it easier to find specific traces by searching through inputs, outputs, and span details. (#18683, @dbczumar)
• 🌐 Gemini TypeScript SDK: Auto-tracing support for Gemini in TypeScript, expanding MLflow's observability capabilities for JavaScript/TypeScript AI applications. (#18207, @joelrobin18)

Breaking Changes

• SQLite as Default Backend: MLflow now uses SQLite as the default backend instead of file-based storage, unless existing mlruns data is detected. This improves performance and reliability for tracking experiments. (#18497, @harupy)