Request for Integration of Mistral Models into Langfuse

[Zherdev1996]

Describe the feature or potential improvement

Our team frequently utilizes Mistral AI models for various natural language processing tasks. We would like to request the addition of support for Mistral models on the Langfuse platform. This would enable us to manage our workflow within Langfuse, leveraging its robust features while utilizing the models we depend on. Pricing info - https://docs.mistral.ai/platform/pricing/

Additional information

No response

[clemra]

Hi @Zherdev1996,

Thank you for suggesting this. We like the mistral models and want to do more to support them natively in the future. For now, here are two ways to work around it:

Cost & Token Usage: You can already add the Mistral Pricing to Langfuse by yourself, have a look at how here: https://langfuse.com/docs/model-usage-and-cost

Mistral SDK: Conveniently, Mistral uses the same underlying logic as OpenAI in their SDK. So you can use the Langfuse OpenAI SDK and exchange the base url to mistral to use it to very conveniently track mistral executions: https://langfuse.com/docs/integrations/openai/get-started

[marcklingen]

We recently added a notebook on how to best wrap the new mistral sdk with the Langfuse Decorator: https://langfuse.com/guides/cookbook/integration_mistral_sdk