It looks like langfuse saves the data from base64 uri to s3 (compatible) and then references the attachment link in the trace to make this work.
This would be awesome if LiteLLM proxy automatically did the work with Langfuse API to upload the image and trace sent to langfuse to get these multimodal traces working with base64 data.
The Feature
https://langfuse.com/docs/tracing-features/multi-modality#api
It looks like langfuse saves the data from base64 uri to s3 (compatible) and then references the attachment link in the trace to make this work. This would be awesome if LiteLLM proxy automatically did the work with Langfuse API to upload the image and trace sent to langfuse to get these multimodal traces working with base64 data.
Motivation, pitch
user request
Twitter / LinkedIn details
No response