[BUG] - OPENAI_API only working with ollama

[hnico21]

First Check

• This is not a feature request.
• I added a very descriptive title to this issue (title field is above this).
• I used the GitHub search to find a similar issue and didn't find it.
• I searched the Mealie documentation, with the integrated search.
• I already read the docs and didn't find an answer.
• This issue can be replicated on the demo site (https://demo.mealie.io/).

What is the issue you are experiencing?

I want to switch the AI. Im currently runnig ollama succsessfully, but I can't get GPT-4 to work. I also tried ollama and openai API provided by my Open-Webui instance (a Webui for ollama and other AIs) .
#4443 has a similar problem.

Steps to Reproduce

      OPENAI_BASE_URL: "http://192.168.0.5:11434/v1"
      OPENAI_API_KEY: "ollama"
#      OPENAI_BASE_URL: http://open-webui:8080/openai
#      OPENAI_API_KEY: "sk-Private"
#      OPENAI_BASE_URL: "https://api.openai.com/v1"
#      OPENAI_API_KEY: "sk-Private"
#      OPENAI_MODEL: "gpt-4"
      OPENAI_MODEL: "llama3.2:latest"
      OPENAI_WORKERS: 1
      OPENAI_ENABLE_IMAGE_SERVICES: false

Please provide relevant logs

relevant Open-Webui logs:

When I make an openai-API call to Open-Webui to send it to Openai

ERROR [open_webui.apps.openai.main] 400, message='Bad Request', url='https://api.openai.com/v1/chat/completions'
Traceback (most recent call last):
  File "/app/backend/open_webui/apps/openai/main.py", line 475, in generate_chat_completion
    r.raise_for_status()
  File "/usr/local/lib/python3.11/site-packages/aiohttp/client_reqrep.py", line 1121, in raise_for_status
    raise ClientResponseError(
aiohttp.client_exceptions.ClientResponseError: 400, message='Bad Request', url='https://api.openai.com/v1/chat/completions'
INFO:     172.27.0.14:53236 - "POST /openai/chat/completions HTTP/1.1" 400 Bad Request

When I make an openai-API call to Open-Webui for other AI (cloudflare AI Worker):
INFO: 172.27.0.14:49148 - "POST /openai/chat/completions HTTP/1.1" 200 OK

When I make an ollama-API call to Open-Webui for ollama
INFO: 172.27.0.14:51222 - "POST /ollama/v1/chat/completions HTTP/1.1" 400 Bad Request

Mealie Version

v2.1.0

Deployment

Docker (Linux)

Additional Deployment Details

running on ubuntu server with mealie and open-webui in docker and ollama on it. also NPM as reverse proxy. Open-Webui-API-Endpoints

[gmag11]

You may try mistral models. They have a generous free tier. It may happen that their OpenAI API is not totally compliant. In that case you can use litellm projects as a proxy for many local (ollama) or cloud models, including mistral.

[michael-genson]

I'm not familiar with Open Web UI, but for the OpenAI integration, you shouldn't provide the base URL to Mealie, you only need to provide the OpenAI API Key

[hnico21]

Thanks for your replies. Apparently I overcomplicated the issue. Just providing the API-Key did the trick.
Now it looks like the issue only occurs when I try to specify "gpt-4" as a model. "gpt-4o" works

      OPENAI_API_KEY: "sk-Private"
      OPENAI_MODEL: "gpt-4"

Apparently there is a problem with cloudflares-API:

{
  "messages": [
    {
      "role": "system",
      "content": "You are a bot that parses user input into recipe ingredients. You will ......."
    },
    {
      "role": "user",
      "content": [
        {
          "type": "text",
          "text": "[\"2 tbsp minced cilantro, leaves and stems\"]"
        }
      ]
    }
  ],
  "return_citations": true,
  "return_images": true
}

Answer HTTP 400:

{
  "errors": [
    {
      "message": "AiError: Bad input: must have required property 'prompt', 'messages1content' must be string, must match exactly one schema in oneOf",
      "code": 5006
    }
  ],
  "success": false,
  "result": {},
  "messages": []
}

I will contact cloudflare for this.

[michael-genson]

Glad you got it working!