ChatGPT API with model gpt-4 is not using GPT4. It's completely different from CHATGPT PLUS GPT4

when I query this

ozgur@Ozgurs-MacBook-Pro ~ % curl https://api.openai.com/v1/chat/completions
-H “Content-Type: application/json”
-H “Authorization: Bearer [REDACTED]”
-d ‘{
“model”: “gpt-4”,
“messages”: [{“role”: “user”, “content”: “What is your model number!”}]
}’
{“id”:“chatcmpl-6wVWzn95O6iDc1Z9W1kJNIZZtZZLh”,“object”:“chat.completion”,“created”:1679402249,“model”:“gpt-4-0314”,“usage”:{“prompt_tokens”:12,“completion_tokens”:45,“total_tokens”:57},“choices”:[{“message”:{“role”:“assistant”,“content”:“As an AI language model, I do not have a model number like a physical device or product would. I am powered by OpenAI’s GPT-3, which stands for Generative Pre-trained Transformer 3.”},“finish_reason”:“stop”,“index”:0}]}

vs CHATGPT PLUS GPT4 says

“My current model is based on the GPT-4 architecture, which is an advanced version of the well-known GPT-3 by OpenAI. As an AI language model, my primary function is to understand and generate human-like text based on the input I receive. Keep in mind that my knowledge is up to date as of September 2021, so I may not be aware of more recent developments or information.”

1 Like

Yes. I have noticed the exact same thing. This is concerning to me…

GPT-4 was created after 2021 which is the training cut off date for both 3.5 and 4, they have a small instruction in their system prompts which tells them what model they are based on, depending how how the model is asked they often do not know. If you make a call to GPT-4 then that is what you get, asking the model directly will not yield a consistent result.