I am not sure, but it does not look like this for all my buttons blocks on the canvas.

I wonder if any of you are using models like Deepseek R1, Sonar Reasoning or o1 through Together and in Typebot? Have you tried? Does it work as well as with Open AI integration? Can you use pre-configured assistants? What are your costs?

weird, isn't it ? what am I supposed to do ?

I ask an OpenAI assistant to generate a list of three subjects. Here's my prompt:

...

Present these 3 topics strictly in a list in json format without any other surrounding text like this:
["Subject 1",Subject 2",Subject 3"].

The result will then be used dynamically to propose a multiple-choice block to the user.

...

Then I save the response in a variable.

But when I reuse the variable to display buttons dynamically afterwards, it doesn't work, I just get a button containing the whole list.

How can I correct this misinterpretation?

How do you promote your bot for those who sell it?

And how do you generate purchase invoices?

What does system, user, assistant or dialogue are for here ? I understand assistant is a call to assistant, but I don't understand other options.

Is it possible to use custom GPT inside Typebot ? If yes, how ? Have some of you ever tried it ?