← Back to changelog
August 21, 2024
Support for Multi-Modal Traces in Langfuse UI
Marlies Mayerhofer
Add multiple modalities to a single trace, including text and images.
Langfuse now supports rendering multi-modal traces. In addition to text, you may now include images (passed via URL). We follow OpenAI's content format convention. For more details, please refer to Multi-Modal Tracing.
How to trace vision input in Langfuse?
If you use the OpenAI Python SDK, you can follow our OpenAI cookbook which includes an example for tracing images in Langfuse.