How to call AzureOpenAI API with PyRIT?

Question

Im triying to do a basic PyRIT project but im not able to use the AzureOpenAI with PyRIT.

I have this code woking, but not with PyRIT:

from openai import AzureOpenAI

client = AzureOpenAI(
api_key="myApiKey",  
azure_endpoint="myEndpoint"
)

try:
   response = client.chat.completions.create(
      model="myModel",
      messages=[
         {"role": "system", "content": "You're a helpful assistant."},
         {"role": "user", "content": "Hello, how are you?"}
      ],
      temperature=0.0
   )
   print(response.choices[0].message.content)
except Exception as e:
   print(f"Error: {e}")

This code works, but now i want to use PyRIT.

This is my PyRIT code:

from pyrit.common import IN_MEMORY, initialize_pyrit
from pyrit.orchestrator import PromptSendingOrchestrator
from pyrit.prompt_target import OpenAIChatTarget

initialize_pyrit(memory_db_type=IN_MEMORY)

prompt = "Hello, how are you?"

target = OpenAIChatTarget(
    endpoint="myEndpoint",
    api_key="myApiKey"
)

orchestrator = PromptSendingOrchestrator(target)

response = await orchestrator.send_prompts_async(prompt_list=[prompt])  # type: ignore
await orchestrator.print_conversations_async()  # type: ignore

This code always return this Exception: Error sending prompt with conversation ID: "anyConversationIDThere."

I tried different ways to connect with the API but always fails the code.

Any idea?

Make sure you are using the correct Azure OpenAI endpoint URL, including your custom subdomain and the appropriate API version like api-version=2023-03-15-preview. — Dasari Kamali
– Dasari Kamali, Commented Apr 1 at 8:48
That was, the endpoint I received was incomplete, thank you very much. — user30129718
– user30129718, Commented Apr 1 at 11:26
@DasariKamali i can´t because its a comment, not an answer, if you want you can leave an answer and i will mark it as answer. Im new to this maybe there is another way to do that. — user30129718
– user30129718, Commented Apr 3 at 6:22

Dasari Kamali · Accepted Answer · 2025-04-03 07:17:02Z

When using Azure OpenAI with PyRIT, always ensure that the endpoint includes the custom Azure OpenAI subdomain and the correct API version.

https://<openaiName>.openai.azure.com/openai/deployments/<deploymentName>/chat/completions?api-version=2023-03-15-preview

I have tried the below code using PyRIT with correct endpoint and successfully got the output response.

Code :

import asyncio
from pyrit.common import IN_MEMORY, initialize_pyrit
from pyrit.orchestrator import PromptSendingOrchestrator
from pyrit.prompt_target import OpenAIChatTarget

initialize_pyrit(memory_db_type=IN_MEMORY)

AZURE_OPENAI_ENDPOINT = "https://<openaiName>.openai.azure.com/"  
DEPLOYMENT_NAME = "<modelName>"  
API_VERSION = "2023-03-15-preview"
API_KEY = "<apiKey>" 

target = OpenAIChatTarget(
    endpoint=f"{AZURE_OPENAI_ENDPOINT}openai/deployments/{DEPLOYMENT_NAME}/chat/completions?api-version={API_VERSION}",
    api_key=API_KEY
)

orchestrator = PromptSendingOrchestrator(target)
async def send_prompt(prompt: str):
    """ Sends the prompt and returns the responses. """
    try:
        return await orchestrator.send_prompts_async(prompt_list=[prompt])  
    except Exception as e:
        print(f"Error sending prompt: {e}")
        return None

def extract_assistant_response(response):
    """ Extracts and prints the assistant's response correctly. """
    if not hasattr(response, "request_pieces"):
        print("Unexpected response format:", response)
        return

    for piece in response.request_pieces:
        if isinstance(piece, dict):
            for key, value in piece.items():
                if isinstance(value, str) and "assistant" in value.lower():
                    print(f"Assistant: {value}")

async def main():
    """ Main function to send prompt and process responses. """
    prompt = "Hello, how are you?"
    responses = await send_prompt(prompt)
    if responses:
        for response in responses:
            print(f"Debug Response: {response.__dict__}") 
            extract_assistant_response(response)
        await orchestrator.print_conversations_async() 
asyncio.run(main())

Output :

I successfully got the output response as shown below.

enter image description here

Collectives™ on Stack Overflow

How to call AzureOpenAI API with PyRIT?

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related