Pydantic Logfire is an observability platform from the team behind Pydantic Validation. It wraps OpenTelemetry with an opinionated Python SDK, collects traces, metrics, and logs, and shows them in a dashboard that you query with standard SQL. The hosted product also covers LLM apps and agents, with distributed tracing, evals, and cost tracking.
You instrument code in two ways: manual spans and log calls with structured arguments, or automatic instrumentation of popular packages such as FastAPI, database connectors, and HTTP libraries. Because the data is OpenTelemetry, existing instrumentation and tooling carry over, and other languages can send data too.
The Python SDK is open source, and SDKs for TypeScript and Rust also exist. These SDKs can export to any OTel-compatible backend. The Logfire platform (the UI and backend) is closed source and runs as a hosted service, with self-hosting available under an enterprise license.
Features
- OpenTelemetry signals: traces, metrics, and logs, built on OpenTelemetry instrumentation
- Python-centric views: rich display of Python objects, event-loop telemetry, and profiling of Python code and database queries
- SQL querying: query telemetry with standard SQL, including from existing BI tools and database libraries
- Package integrations: automatic instrumentation for frameworks such as FastAPI plus database and HTTP libraries
- Pydantic analytics: inspects data flowing through Pydantic Validation models and reports on validations
- AI observability: tracing of LLM apps and agents with evals and cost tracking
- Multi-language SDKs: open-source SDKs for Python, TypeScript, and Rust that export to any OTel-compatible backend
