o1, o1-preview o1-mini | tier limited

[appatalks]

New models o1-preview o1-mini made avail!

Just now gotta figure out they API calls.

Error 400: Invalid Request
{
  "error": {
    "message": "Unsupported value: 'messages[0].role' does not support 'system' with this model.",
    "type": "invalid_request_error",
    "param": "messages[0].role",
    "code": "unsupported_value"
  }
}

[appatalks]

• https://platform.openai.com/docs/models#o1

| Model                  | Context window | Max output tokens | Training data  | 
|------------------------|----------------|-------------------|----------------| 
| o1-preview             | 128,000 tokens | 32,768 tokens     | Up to Oct 2023 | 
| o1-preview-2024-09-12  | 128,000 tokens | 32,768 tokens     | Up to Oct 2023 | 
| o1-mini                | 128,000 tokens | 65,536 tokens     | Up to Oct 2023 | 
| o1-mini-2024-09-12     | 128,000 tokens | 65,536 tokens     | Up to Oct 2023 | 

[appatalks]

"message": "Unsupported value: 'messages[0].role' does not support 'system' with this model.",

• https://platform.openai.com/docs/api-reference/chat/create

  model="o1-preview",
  messages=[
    {"role": "system", "content": "You are a helpful assistant."},
    {"role": "user", "content": "Hello!"}
  ]

Well that's odd...
API doc's don't show a structural difference between o1 and other models... (see further down below)

Quick Start Guide is different:

    model="o1-preview",
    messages=[
        {
            "role": "user", 
            "content": "Write a bash script that takes a matrix represented as a string with format '[1,2],[3,4],[5,6]' and prints the transpose in the same format."
        }
    ]

Other Gotcha's:

• max_tokens Deprecated

  • This value is now deprecated in favor of max_completion_tokens, and is not compatible with o1 series models.

• Modalities: text only, images are not supported.

• Message types: user and assistant messages only, system messages are not supported. <---- Well there's your problem.

• Tools: tools, function calling, and response format parameters are not supported.

• Logprobs: not supported.

• Other: temperature and top_p are fixed at 1, while presence_penalty and frequency_penalty are fixed at 0.

• Assistants and Batch: these models are not supported in the Assistants API or Batch API.

[appatalks]

Maybe I can introduce an if statement here while we are in beta to excluded system from the call, and having sModel === "o1-preview|o1-mini" or something like that.,

      const iMessages = [
        { role: 'system', content: 'You are Eva. You have access to previous chats and responses. You have access to real-time news, information and media. You will keep conversation to a minimum and answer to the best of your abilities. When you are asked to show an image, instead describe the image with [Image of <Description>].' },
        { role: 'user', content: selPers.value + " " + dateContents },
      ];

      // Store the initial messages in localStorage
      localStorage.setItem("messages", JSON.stringify(iMessages));
    }

Also need to update the max_tokens parameter to be max_completion_tokens

[appatalks]

• Updated Guidance

Developer messages are the new system messages: Starting with o1-2024-12-17, o1 models support developer messages rather than system messages Learn More

Markdown formatting: Starting with o1-2024-12-17, o1 models in the API will avoid generating responses with markdown formatting. To signal to the model when you do want markdown formatting in the response, include the string Formatting re-enabled on the first line of your developer message.

const response = await openai.chat.completions.create({
  model: "gpt-4o",
  messages: [
    {
      "role": "developer",
      "content": [
        {
          "type": "text",
          "text": `
            You are a helpful assistant that answers programming 
            questions in the style of a southern belle from the 
            southeast United States.
          `
        }
      ]
    },
    {
      "role": "user",
      "content": [
        {
          "type": "text",
          "text": "Are semicolons optional in JavaScript?"
        }
      ]
    }
  ]
});

[appatalks]

Blocker for o1

Current o1 models
Available to Tier 5 customers only

Tier 5 $1,000 paid and 30+ days since first successful payment $200,000 / month

[appatalks]

• max_tokens no longer used. See PR Adding o1 model and updating endpoint usage #64
• Developer messages are the new system messages See PR Adding o1 model and updating endpoint usage #64

: Starting with o1-2024-12-17, o1 models support developer messages rather than system messages

Going to wait a bit longer to see if I still have to add a new if block just just for o1-* models.

• Edit See o1-preview and o1-mini #67