Adding custom model definitions

[PrinceBaghel258025]

• Allow to define custom model definition

[Steffen911]

@PrinceBaghel258025 Do I understand your request correctly that you would want to configure Langfuse Custom Model definitions on the helm chart? So far we only support them through the UI or the API (https://langfuse.com/docs/model-usage-and-cost#custom-model-definitions). Doing it on the helm chart would require an environment based initialization. I.e. if I understand you correctly this feature request would be better placed in our main repository (https://github.com/langfuse/langfuse).

[PrinceBaghel258025]

@PrinceBaghel258025 Do I understand your request correctly that you would want to configure Langfuse Custom Model definitions on the helm chart? So far we only support them through the UI or the API (https://langfuse.com/docs/model-usage-and-cost#custom-model-definitions). Doing it on the helm chart would require an environment based initialization. I.e. if I understand you correctly this feature request would be better placed in our main repository (https://github.com/langfuse/langfuse).

Thanks,
I was using langfuse and litellm together. As to get benefit call failures, falling back to models from different provider etc.
as we have various aliases for underlying models.
Just a question, does these custom model definitions are being saved to postgres ? If so, we don't want bother anyone, it's just one time effort

[PrinceBaghel258025]

Is there also a way to enable beta features through this helm chart ? like evaluations, post-hog integration etc

[marcklingen]

Is there also a way to enable beta features through this helm chart ? like evaluations, post-hog integration etc

these are not yet available for self-hosting, will be part of v3