Workflow LLM not working

[humboldthills]

I have entered the API key and URL for OpenAI, it gets verified and I have tried many of the model options from the drop-down but no matter what I choose I get: The model us.anthropic.claude-sonnet-4-20250514-v1:0 does not exist or you do not have access to it.', 'type': 'invalid_request_error', 'param': None, 'code': 'model_not_found' in the logs.

So I tried to verify my Anthropic API Key to choose one of their models. It will not verify with https://api.anthropic.com https://api.anthropic.com/v1 or https://api.anthropic.com/v1/. It always fails and the in the logs when trying to verify the Anthropic API key it seems to be checking openai:

ERROR    | llm_api.py:verify_openai_key:115 | API validation failed with status code: 404
2026-04-18 02:54:16 | ERROR    | llm_api.py:verify_openai_key:115 | API validation failed with status code: 401
2026-04-18 02:55:04 | ERROR    | llm_api.py:verify_openai_key:115 | API validation failed with status code: 404

That is trying to authenticate Anthropic to get around the error in using OpenAI. I will share some output from when I try to use OpenAI since it does authenticate. The chat works fine but as soon as it tries to spin up an agent the log shows:

2026-04-18 03:14:33 | ERROR    | mcp_client.py:comfyui_agent_invoke:503 | Non-retryable streaming error or max retries reached: Error code: 404 - {'error': {'message': 'The model `us.anthropic.claude-sonnet-4-20250514-v1:0` does not exist or you do not have access to it.', 'type': 'invalid_request_error', 'param': None, 'code': 'model_not_found'}}
2026-04-18 03:14:33 | ERROR    | mcp_client.py:comfyui_agent_invoke:504 | Traceback: Traceback (most recent call last):
  File "/workspace/ComfyUI/custom_nodes/ComfyUI-Copilot/backend/service/mcp_client.py", line 460, in comfyui_agent_invoke
    async for stream_data in process_stream_events(result):
  File "/workspace/ComfyUI/custom_nodes/ComfyUI-Copilot/backend/service/mcp_client.py", line 455, in process_stream_events
    raise e
  File "/workspace/ComfyUI/custom_nodes/ComfyUI-Copilot/backend/service/mcp_client.py", line 338, in process_stream_events
    async for event in stream_result.stream_events():
  File "/workspace/miniconda3/envs/comfyui/lib/python3.11/site-packages/agents/result.py", line 773, in stream_events
    raise self._stored_exception
  File "/workspace/miniconda3/envs/comfyui/lib/python3.11/site-packages/agents/run_internal/run_loop.py", line 1012, in start_streaming
    turn_result = await run_single_turn_streamed(
                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/workspace/miniconda3/envs/comfyui/lib/python3.11/site-packages/agents/run_internal/run_loop.py", line 1473, in run_single_turn_streamed
    async for event in retry_stream:
  File "/workspace/miniconda3/envs/comfyui/lib/python3.11/site-packages/agents/run_internal/model_retry.py", line 652, in stream_response_with_retry
    event = await stream.__anext__()
            ^^^^^^^^^^^^^^^^^^^^^^^^
  File "/workspace/miniconda3/envs/comfyui/lib/python3.11/site-packages/agents/models/openai_chatcompletions.py", line 248, in stream_response
    response, stream = await self._fetch_response(
                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/workspace/miniconda3/envs/comfyui/lib/python3.11/site-packages/agents/models/openai_chatcompletions.py", line 441, in _fetch_response
    ret = await self._get_client().chat.completions.create(**create_kwargs)
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/workspace/miniconda3/envs/comfyui/lib/python3.11/site-packages/openai/resources/chat/completions/completions.py", line 2714, in create
    return await self._post(
           ^^^^^^^^^^^^^^^^^
  File "/workspace/miniconda3/envs/comfyui/lib/python3.11/site-packages/openai/_base_client.py", line 1913, in post
    return await self.request(cast_to, opts, stream=stream, stream_cls=stream_cls)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/workspace/miniconda3/envs/comfyui/lib/python3.11/site-packages/openai/_base_client.py", line 1698, in request
    raise self._make_status_error_from_response(err.response) from None
openai.NotFoundError: Error code: 404 - {'error': {'message': 'The model `us.anthropic.claude-sonnet-4-20250514-v1:0` does not exist or you do not have access to it.', 'type': 'invalid_request_error', 'param': None, 'code': 'model_not_found'}}

2026-04-18 03:14:33 | INFO     | mcp_client.py:comfyui_agent_invoke:530 | Total tool results: 0
2026-04-18 03:14:33 | INFO     | mcp_client.py:comfyui_agent_invoke:540 | === End Tool Results Summary ===

2026-04-18 03:14:33 | INFO     | conversation_api.py:invoke_chat:310 | -- Received ext data: None, finished: True
2026-04-18 03:14:33 | INFO     | conversation_api.py:invoke_chat:335 | -- Sending final response: 44 chars, ext: False, finished: True

This seems like the same behavior seen in Issue #116 but I am still seeing the broken behavior unless I am missing something in the config/setup.

I can provide more info if it would help.