I am not sure what you mean by "difference between regression and separate location model at parametric survival fit." I will use an example from the Reliability sample data folder, Motorette. I fit a life model using Temp as the factor for both the location and the scale parameter. All terms are significant.

So the location and the scale parameter varies with temperature. Examine the Parameter Estimates in particular.

How are these parameter estimates used? What does the model look like? I saved the prediction formulas for the probability and the 0.1 quantile. Here they are:

(Predicted Probability)

15.965159157 - 0.043882405* Temp is the location parameter. 2.039358492 - 0.008725175*Temp is the scale parameter.

(Predicted 0.1 Quantile)

The interpretation for the Weibull parameters based on the linear predictors above is the same for the second formula.

Learn it once, use it forever!