Parallel tool/function calling is not supported

[Puzer]

The provided code snippet works fine if LLM (v3-5-sonnet) decides to use only one function, but failing if LLM is calling two functions.
This snippet also works fine if I use gpt-4o model for parallel function calls.

import json
from langchain_core.tools import tool
from langchain_core.utils.function_calling import convert_to_openai_tool
from openai import AzureOpenAI

query = "What is 3 * 12 and 6 * 2?" # parallel FC use-case <- failing
# query = "What is 3 * 12?" # a regular FC <- works fine

@tool
def add(a: int, b: int) -> int:
    """Add two integers.

    Args:
        a: First integer
        b: Second integer
    """
    return a + b

@tool
def multiply(a: int, b: int) -> int:
    """Multiply two integers.

    Args:
        a: First integer
        b: Second integer
    """
    return a * b

tools = [add, multiply]
available_functions = {t.name:t.func for t in tools}
tools_openai_schema = [convert_to_openai_tool(t) for t in tools]

messages = [{"role": "user", "content": query}]


client = AzureOpenAI(
    api_key="XXX",  
    api_version="2023-03-15-preview",
    azure_endpoint="XXX"
)

model_name = "anthropic.claude-v3-5-sonnet"
completion = client.chat.completions.create(
    model=model_name,
    messages=messages,
    tools=tools_openai_schema
)

response_message = completion.choices[0].message
tool_calls = response_message.tool_calls
assert tool_calls

response_message.content = None# instead of empty string (raises error) convert that to None
messages.append(response_message.to_dict())


# Parallel function calling
# https://platform.openai.com/docs/guides/function-calling/configuring-parallel-function-calling

for tool_call in tool_calls:
    function_name = tool_call.function.name
    function_to_call = available_functions[function_name]
    function_args = json.loads(tool_call.function.arguments)
    function_response = function_to_call(**function_args)

    message_to_append = {
            "tool_call_id": tool_call.id,
            "role": "tool",
            "name": function_name,
            "content": str(function_response),
        }
    messages.append(message_to_append)  # extend conversation with function response, as in docs

completion = client.chat.completions.create(
    model=model_name,
    messages=messages,
    tools=tools_openai_schema
)

print(completion.choices[0].message.content)

I'm getting error:

Error code: 400 - {'error': {'message': 'Error code: 400 - {\'message\': "messages.2: the following `tool_use` ids were not found in `tool_result` blocks: {\'...\'}"}', 'type': 'invalid_request_error', 'code': '400'}}

[Puzer]

Seems that I tested outdated version of adapter.
Issue resolved some time ago.

If Sonnet V2 is not using functions in parallel and uses them sequentially, one-by-one, but you would like to have them in parallel (in one LLM call), just add to the System Message something like "Use function calls in parallel whereas possible".