MistralAPIStatusException: Cannot stream response. Status: 429

[ttkrpink]

I put one training question/sql pair into the model.
I trained the model using vn.train().
The vn.ask return with 429 error.

[zainhoda]

Do you have a Mistral account? Currently the configuration for Mistral requires a Mistral API key

[ttkrpink]

I did have the API key.

API_Mistral = "xxxxxxxxxxxxxxxxxxxxxx"
API_Vanna ="xxxxxxxxxxxxxxxxxxxxxx"

from vanna.chromadb.chromadb_vector import ChromaDB_VectorStore
from vanna.mistral.mistral import Mistral

Database = './database.sqlite'
class MyVanna(ChromaDB_VectorStore, Mistral):
    def __init__(self, config=None):
        ChromaDB_VectorStore.__init__(self, config=config)
        Mistral.__init__(self, config={'api_key': API_Mistral, 'model': 'mistral-tiny'})

vn = MyVanna()

[zainhoda]

Closing this issue as non-reproducible. Our integration test runs against Mistral on every merge and nobody else has reported this issue:

https://github.com/vanna-ai/vanna/actions/runs/8180364867/job/22368242743#step:5:35

[varunsingh3000]

This happened to me because my monthly usage limit had reached on the platform. Increasing the limit fixed it. But if this was the case, then I think the error description needs to be more clear and explicit regarding what caused it.

[zainhoda]

Thanks @varunsingh3000 and @ttkrpink -- This type of error comes directly from the API provider (Mistral). I think the right thing to do is to keep the raw error as is. It's really Mistral that should provide more context about the error via their package: https://github.com/mistralai/client-python

I think once we get into the business of "interpreting" API errors, it's going to get unmanageable for us very quickly, especially if the underlying Mistral package changes its error handling/messages.