Which model does OpenAI’s newly released code interpreter use? Is it GPT4 or a variant? Is it available via API?

Welcome to the forum! It uses GPT-4 and had a number of support functions around it, you can build agents to perform similar tasks via the API.

I wonder if someone can help me figure out why I am seeing chatgpt-3 as the model when using code interpreter. I realized this when the model refused to do sentiment analysis because it couldn’t. I start a new chat on chat opena .com and pick gpt-4 and then code interpreter. I then ask what model …

The training data has a cut-off date of 2021, GPT-4 was built after that point, thus no reference to it, asking the model what version it is is unreliable.

If you really want to know, banana test it. “Today I have four banana. Yesterday, I ate two. How many bananas do I have now?” GPT-4 can get this right.

Interesting Sentiment Analysis can be found here Google Cloud Natural Language offer it too . I’ve found it’s worth using a mix of Apis in a single system. NER from one, Keywords from another and a couple of Summaries etc. Keeps them on their toes :wink:

I’m sorry, I wasn’t as clear as I should have been. My issue wasn’t that it reports as 3 vs 4. That’s just showing what I was facing. The real issue is that it refuses to do sentiment analysis because it is gtp3. That’s my issue and I only touched on it briefly in my previous post, so sorry about t…

I love this test. I opened a new code interpreter chat and typed exactly what you said. Here is the response: You have four bananas today and you ate two yesterday. The bananas you ate yesterday would not affect the number of bananas you have today. So, you still have four bananas today. So I thi…

What was the prompting that lead up to that reply, also the intermediate replies, do you have a chat link you would be willing to share?

I just started a new chat, loaded my csv file and asked if the following and then got the response below: Please do sentiment analysis of this data … Sentiment analysis is typically done using a pre-trained model. One common approach is to use models such as Vader from NLTK (Natural Language Tool…

That’s the correct answer: GPT-4 GPT-3.5 will write an incorrect answer: that the bananas you have today are then modified by the amount eaten yesterday. You got outsmarted by a computer.

Which model does code interpreter use?

API

Topic		Replies	Views
Confusing response regarding what model used API gpt-4	4	1828	October 11, 2023
How to access GPT4 with Code Interepreter via API? API gpt-4	3	3610	December 17, 2023
GPT-4 API behaves likes it's GPT-3 API	13	5746	December 17, 2023
When calling the API, the model passed in is gpt-4-0314. When asked what language model it is, it still says that it is gpt-3. Why? API gpt-4	16	2725	June 13, 2023
API model set to GPT-4 but seems to respond as GPT-3? API	14	7712	November 28, 2023

Which model does code interpreter use?

Related topics