vertexai: Add "mistral-large-2411@001" to model garden maas

[fjeanchar2]

Description :

🆕 New Feature : Add Mistral's latest model"mistral-large-2411@001" to model garden maas.

Tests :

✅ Test

from langchain_google_vertexai.model_garden_maas.mistral import VertexModelGardenMistral
 
model = VertexModelGardenMistral(
    model_name="mistral-large-2411@001",
    max_output_tokens=8000,
    top_k=10,
    temperature=0.0,
    location="europe-west4",
    project="sandbox-fjeancharles",
)

model.invoke("hello")
output: AIMessage(content="Hello! How can I assist you today? Let's have a friendly conversation. 😊 How are you doing?", additional_kwargs={}, response_metadata={'token_usage': {'prompt_tokens': 5, 'total_tokens': 29, 'completion_tokens': 24}, 'model': 'mistral-large-2411@001', 'finish_reason': 'stop'}, id='run-3ab72497-c2eb-44e0-bd34-ddefeb315bf8-0', usage_metadata={'input_tokens': 5, 'output_tokens': 24, 'total_tokens': 29})

[lkuligin]

integration tests are failing unfortunately. are you sure the model has the same output format as the previous one?

[fjeanchar2]

integration tests are failing unfortunately. are you sure the model has the same output format as the previous one?

I haven't tested the streaming part, but I'll look into it! I let you know

[fjeanchar2]

integration tests are failing unfortunately. are you sure the model has the same output format as the previous one?

@lkuligin a fix has been made in version 0.2.4 of the langchain-mistralai package in the chat_models.py script (L600 and L626) concerning the error found in the stream and astream method (see test details).
https://github.com/langchain-ai/langchain/blob/master/libs/partners/mistralai/langchain_mistralai/chat_models.py

I can see in the poetry.lock file that you are with the 0.2.3 of langchain-mistralai package.