Error When pinging chatgpt-4o-latest model

API Bugs
1

Hello everyone,

I’m encountering a strange issue and hope someone can help me out. Whenever I try to ping the chatgpt-4o-latest model, I consistently receive the following error:

[2024-09-23 04:52:42] [IP: xxxx:xxxx:xxxx:xxxx:xxxx:xxxx:xxxx:xxxx] {
    "error": {
        "message": "The server had an error processing your request. Sorry about that! You can retry your request, or contact us through our help center at help.openai.com if you keep seeing this error. (Please include the request ID req_f3e243fb6eb05a42a1122c1d6575b53e in your email.)",
        "type": "server_error",
        "param": null,
        "code": null
    }
}

Interestingly, this error only occurs with the chatgpt-4o-latest model. When I use the regular model, everything works fine.

It also does not happen in the playground.

My questions are:

  1. Has anyone experienced something similar?
  2. Could this be related to my configuration or access settings?
  3. Are there any known issues with the Latest-Model that I should be aware of?

I would appreciate any help or insights. If additional information is needed, please let me know.

Thank you in advance!
Michael

2

“pinging” seems to be working…output from the model

To send an ICMP echo request (ping) to the IP address 192.168.20.40 with TTL 10, I’ll use the ping command in the terminal.

Let’s proceed:

ping -c 4 -t 10 192.168.20.40

{‘prompt_tokens’: 193, ‘completion_tokens’: 64, ‘total_tokens’: 257, ‘completion_tokens_details’: {‘reasoning_tokens’: 0}}

Keep in mind this is a special trained model that doesn’t accept tools (like ChatGPT does) specifically so that you can’t create your own bot that has a code interpreter, for example. Nor will it emit the correct tokens for tools, because they switch training weights or logits on tools vs no tools submodels. Making the model of little value.

The wrong kind of input or restricted max_tokens could have it produce a token sequence that caused an internal parse error.

Start with:

chat = [{"role": "system","content": [{"type": "text", "text":
"""You are ChatGPT, a large language model trained by OpenAI, based on the GPT-4 architecture.
Knowledge cutoff: 2023-04
Current date: 2024-09-30

Image input capabilities: Enabled
Markdown renderer: Disabled
3

I didn’t really mean to send a ping in the literal way, but a curl request, like in my example.

4

While I literally gave an in-context tool definition to the ChatGPT model providing it a root Debian Linux bash environment with internet access, and more to overcome its refusals about not doing internet stuff, up to where it couldn’t actually emit the correct output sequence to a tool recipient when writing its output, the main point to glean:

  • Up for me, must be you and what you are sending