Nice, can it also trace tts stt vision ?

[ptempier]

Hello

Thanks for the pipeline, i was looking for some better usage stats that what open-webi provides by default.
Using the pipeline with your script, i get some better usage infos, and the users are mapper properly.

I did some tests and tts stt and vision request don't appear in langfuse.
Maybe that's a langfuse, openwebui or pipeline limitation.
But if it's not , could it be possible to add theses to the pipeline ?

[karaketir16]

Hi @ptempier ,

I can see the image inputs from webui on Langfuse, although I haven't tested other inputs yet, so I can't confirm whether they are supported or not.

This example functions as a pipeline filter. You can check the logs for "inlet" — if you see the logs when using these inputs, then it's likely that Pipelines supports them. After that, you can review the Langfuse documentation to verify if there is support for these inputs and modify the filter accordingly.

I'd be happy to have your contributions if you manage to get this working!