Chat Completions API Documentation

This endpoint is used to provide a prompt, and the model will return one or more predicted completions. You can generate text based on the conversation history and control the randomness, length, and format of the generated output.

• Endpoint Name: List Models / Chat Completions
• HTTP Method: GET (Note: Based on the provided data definition)
• Endpoint URL: https://api.codingplanx.ai/v1/models
• Status: Released

---

1. Request Parameters

1.1 Header Parameters

1.2 Body Parameters

Note: Although originally defined as a GET request, based on typical business logic, the following parameters are usually passed via the JSON Body.

---

2. Response Parameters

---

3. Request Examples

Successful Request Example

curl --location --request GET 'https://api.codingplanx.ai/v1/models' \
--header 'Content-Type: application/json' \
--header 'Accept: application/json' \
--header 'Authorization: Bearer YOUR_API_KEY' \
--data '{
  "model": "gpt-3.5-turbo",
  "messages": [
    {
      "role": "user",
      "content": "Hello there, how may I assist you today?"
    }
  ],
  "temperature": 1,
  "top_p": 1,
  "n": 1,
  "stream": false,
  "user": "user-1234"
}'

Successful Response Example

{
    "id": "chatcmpl-123",
    "object": "chat.completion",
    "created": 1677652288,
    "choices": [
        {
            "index": 0,
            "message": {
                "role": "assistant",
                "content": "\r
\r
Hello there, how may I assist you today?"
            },
            "finish_reason": "stop"
        }
    ],
    "usage": {
        "prompt_tokens": 9,
        "completion_tokens": 12,
        "total_tokens": 21
    }
}

---

4. FAQs (Frequently Asked Questions)

Q: Why does the documentation show a GET method, but it looks like a POST request? A: In standard OpenAI specifications, creating completions generally uses the POST method. However, based on the provided metadata, this endpoint is defined as GET. Please verify the actual server-side configuration when making API calls. If GET does not work, try using POST.

Q: How should I choose between temperature and top_p? A: We generally recommend altering either temperature or top_p, but not both. If you want precise and deterministic answers, lower the temperature (e.g., 0.2). If you want more diverse and creative responses, increase it.

Q: How do I enable JSON mode? A: You need to set { "type": "json_object" } within the response_format parameter, and explicitly instruct the model (via System or User messages) to output in JSON format. Otherwise, the model may generate parsing errors or an endless stream of whitespace.

Q: What is the difference between presence_penalty and frequency_penalty? A:

• presence_penalty: Focuses on penalizing topics that have already appeared, encouraging the model to discuss new topics.
• frequency_penalty: Focuses on penalizing exact words/phrases that appear frequently, preventing the model from repeating the same sentences within a short period.

Q: How does streaming output (stream: true) work? A: When streaming is enabled, the API returns data word-by-word via Server-Sent Events (SSE). Each data chunk is a JSON string prefixed with data: , and the stream is terminated with a data: [DONE] message. This is highly useful for displaying long-form text generation in real-time on UI/UX interfaces.