chatcmpl-claude.py

from agents import Agent, ModelSettings, Runner, function_tool, OpenAIChatCompletionsModel
from openai import AsyncOpenAI
import os

custom_client = AsyncOpenAI(
    base_url="https://api.anthropic.com/v1/",
    api_key=os.getenv("ANTHROPIC_API_KEY")
)

my_model = OpenAIChatCompletionsModel(
        model="claude-sonnet-4-5",
        openai_client=custom_client,
    )


@function_tool
def echo_tool(input: str) -> str:
    print("called echo_tool")
    return input

agent = Agent(
    name="Assistant",
    instructions="You are a helpful assistant",
    model=my_model,
    tools=[echo_tool],
    model_settings=ModelSettings(
        extra_body={
            "thinking": { "type": "enabled", "budget_tokens": 2000 }
        }
    )
)

result = Runner.run_sync(
    agent, "Use the echo_tool to echo the message 'Hello, world!"
)
print(result.final_output)
print("------")
print(result.to_input_list())


# openai.BadRequestError: Error code: 400 - {'error': 
#   {'code': 'invalid_request_error', 'message': 
# 'messages.1.content.0.type: Expected `thinking` or `redacted_thinking`,
#  but found `tool_use`. When `thinking` is enabled, a final `assistant` message 
# must start with a thinking block (preceeding the lastmost set of `tool_use` and
#  `tool_result` blocks). We recommend you include thinking blocks from previous turns. 
# To avoid this requirement, disable `thinking`. Please consult our
#  documentation at https://docs.claude.com/en/docs/build-with-claude/extended-thinking', 
# 'type': 'invalid_request_error', 'param': None}}

# 2025/12/6
# It seems official Claude endpoint does not output thinking part.
# So there is no solution.