Your question is not clear...a model is a mathematical equation...a graph is a chart. Can you tell us which you really want?
Sorry for the confusion, I would like to create the response surface model which means a mathematical equation that fits my data using linear least-squares regression.
At the highest level then you'll be accessing one or more of the modeling platforms available in JMP or JMP Pro. There is no 'one size fits all' modeling methodology. Modeling methods have been developed over the years for addressing specific types of practical problems, means by which the data was collected (DOE, happenstance, or some other method?) etc. In general each of these platforms requires you to articulate responses, a model specification, and often times, a specific modeling method. Each platform will produce a model and lots of diagnostics...you get to decide if it's useful for your practical problem. From the sounds of it though you've expressed a desire to use the Fit Model -> Standard Least Squares personality. But there are others too like the general linear models personality that might work too.
I suggest taking a look at the JMP online documentation around Fitting Linear Models to start to get a sense for the best way to proceed: