API Key error

[BodapatiNirupamasai]

litellm.exceptions.AuthenticationError: litellm.AuthenticationError: AuthenticationError: OpenAIException - The api_key client option must be set either by passing api_key to the client or by setting the OPENAI_API_KEY environment variable

[Tlaloc-Es]

Same here

from langchain_openai import AzureChatOpenAI

azure_chatgpt = AzureChatOpenAI(
    azure_deployment="****",
    api_version="****",
    api_key="****",
    azure_endpoint="https://****.openai.azure.com",    
)

This works perfectly, but when try to set that azue_chatgpt to a agent I got the same error

litellm.exceptions.AuthenticationError: litellm.AuthenticationError: AuthenticationError: OpenAIException - The api_key client option must be set either by passing api_key to the client or by setting the OPENAI_API_KEY environment variable

[jamil-z]

Hello! I don't know if anyone was able to figure out how to connect with CrewAI agents and tools for current versions, like 0.7 and above.

I saw that Azure only works for CrewAI 0.11.

[rao208]

from crewai import LLM
import os

os.environ["OTEL_SDK_DISABLED"] = "true"

without setting the "OTEL_SDK_DISABLED" to true I was getting the below error

"requests.exceptions.ConnectTimeout: HTTPConnectionPool(host='telemetry.crewai.com', port=4318): Max retries
exceeded with url: /v1/traces (Caused by ConnectTimeoutError(<urllib3.connection.HTTPConnection object at
0x7f34caf38280>, 'Connection to telemetry.crewai.com timed out. (connect timeout=10)')"

config = configparser.ConfigParser()
config.read("config.ini")

OpenAI35kTurbo = LLM(
model=f"azure/{config["AZURE_OPEN_AI"]["DEPLOYMENT_NAME"]}",
api_version=config["AZURE_OPEN_AI"]["AZURE_OPENAI_API_VERSION"],
api_key=config["AZURE_OPEN_AI"]["AZURE_OPENAI_API_KEY"],
base_url= config["AZURE_OPEN_AI"]["BASE_URL"]
)

Your base_url will be https://${INSTANCE_NAME}.openai.azure.com. Basically it is expecting an endpoint for base_url parameter.

This works perfectly well for me.

Please notice azure in model=f"azure/{config["AZURE_OPEN_AI"]["DEPLOYMENT_NAME"]}". That is important.

[topratiksharma]

This below worked for me as well.

latest crew-ai doesn't have correct details for the setup, the below setup works for me.

import os
from crewai import LLM

def getLLM():
    return LLM(
        model="azure/modelname",
        api_version="version",
        api_key="key",
        base_url="base-url"
    )

[pranjalagg]

I am encountering the same error. I am trying to use Google Gemini, and the above methods do not work for me. I would appreciate any help. Thanks!

[topratiksharma]

I am encountering the same error. I am trying to use Google Gemini, and the above methods do not work for me. I would appreciate any help. Thanks!

Can you post your error message, Along with your setup. It will help looking into the issue.

[pranjalagg]

Sure,

LLM:
llm = ChatGoogleGenerativeAI(model="gemini-1.5-flash",google_api_key=GEM_API)

Agent:

agent1 = Agent(
    role="Communication Analyst",
    goal="Understand the definitions of the classes provided and make a rational decision on how to classify any conversation segment into classes",
    backstory=AGENT1_BACKSTORY,
    LLM=llm,
    api_key=GEM_API
)

Output when the above runs: Checking env var OPENAI_API_KEY: None

... code continues ...

crew.kickoff()
Output:
ERROR:root:LiteLLM call failed: litellm.AuthenticationError: AuthenticationError: OpenAIException - The api_key client option must be set either by passing api_key to the client or by setting the OPENAI_API_KEY environment variable
WARNING:opentelemetry.trace:Overriding of current TracerProvider is not allowed

[Yash-Bavishi]

@pranjalagg

llm attribute of Agent class expects a object of LLM class from crewai rather than langchain_google_genai

I would request you to try the below code

# Imports LLM class from crewai
from crewai import LLM, Agent

# LLM Object from crewai package
llm = LLM(mode="gemini-1.5-flash",api_key=GEM_API)

# create your agent
agent1 = Agent(
    role="Communication Analyst",
    goal="Understand the definitions of the classes provided and make a rational decision on how to classify any conversation segment into classes",
    backstory=AGENT1_BACKSTORY,
    llm=llm,
    api_key=GEM_API
)