/Pydantic

The Eval has landed 🦅

Samuel Colvin avatar
Samuel Colvin
4 mins

We've had an incredibly productive first quarter of 2025, with significant feature enhancements across our entire ecosystem. Here's a comprehensive overview of what we've shipped:

Our new open-source evaluation framework helps you benchmark and evaluate AI models with optional Logfire integration, making it easier to measure and improve LLM performance.

Connect your PydanticAI agents directly to your Logfire data through our new Model Context Protocol integration, enabling more intelligent agents with access to your observability data.

We've expanded our SDK offerings with:


We've launched both Self-Hosted and Enterprise Cloud editions of Logfire for organizations with advanced requirements, offering greater control and customization.

Store and process your observability data entirely within the European Union with our new regional deployment, helping meet data sovereignty and compliance requirements.

Logfire is now SOC2 and HIPAA compliant, making it suitable for use in regulated industries including healthcare.


  • Google Login support for simpler Logfire account creation
  • Enhanced PydanticAI LLM tracing with token counts and tool call details
  • Improved UX for LLM monitoring
  • OpenAI agents instrumentation support
  • OTel logs endpoint for unified observability
  • Significant query speed optimizations
  • Overall user experience enhancements

The latest Pydantic release delivers major performance improvements and new features that enhance the developer experience.

Improvements to Logfire's English-to-SQL AI functionality for more natural data querying.


For more details on any of these updates, please get in touch. As always, we welcome your feedback as we continue to improve our products.